Court of Justice of the European Communities (including Court of First Instance Decisions)

JUDGMENT OF THE GENERAL COURT (Tenth Chamber, Extended Composition)

18 September 2024 (*)

( Competition – Abuse of dominant position – Market for online search advertising intermediation in the EEA – Decision finding an infringement of Article 102 TFEU and of Article 54 of the EEA Agreement – Exclusive supply obligation – Contractual restrictions )

In Case T‑334/19,

Google LLC, established in Mountain View, California (United States),

Alphabet Inc., established in Mountain View,

represented by C. Jeffs, lawyer, J. Holmes KC, and J. Williams, Barrister,

applicants,

supported by

Surfboard Holding BV, established in Zeist (Netherlands), represented by E. Batchelor, Solicitor, and G. de Vasconcelos Lopes, lawyer,

and by

Vinden.NL BV, established in Rijseen (Netherlands), represented by B. Nijhof and N. Strous, lawyers,

interveners,

v

European Commission, represented by N. Khan, A. Dawes, T. Franchoo and C. Urraca Caviedes, acting as Agents,

defendant,

THE GENERAL COURT (Tenth Chamber, Extended Composition),

composed, at the time of the deliberations, of A. Kornezov, President, E. Buttigieg, K. Kowalik-Bańczyk (Rapporteur), G. Hesse and D. Petrlík, Judges,

Registrar: I. Kurme, Administrator,

having regard to the written part of the procedure,

further to the hearing on 2, 3 and 4 May 2022,

gives the following

Judgment

1        By their action based on Article 263 TFEU, the applicants, Google LLC, formerly Google Inc., and its parent company, Alphabet Inc. (together, ‘Google’), seek, primarily, annulment of Commission Decision C(2019) 2173 final of 20 March 2019 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.40411 – Google Search (AdSense)) (‘the contested decision’), or, in the alternative, annulment or reduction of the amount of the fine imposed in that decision.

I.      Background to the dispute

2        Google is an undertaking in the information and communications technology sector which specialises in internet-related products and services and is active within the European Economic Area (EEA). It is known in particular for its general search engine, which allows users to find and access, with their browser of choice and by means of hyperlinks, websites meeting their needs.

3        Google’s search engine, accessible at the address ‘www.google.com’, or at similar addresses with a national extension, yields search results presented on pages appearing on users’ screens. Those results are selected by the search engine either according to general criteria and without the sites to which they link paying Google to appear or according to a specialised logic for the particular type of search carried out; those latter results may also be independent of payments from the websites to which they link.

4        Google’s results pages also include results linked to payments from the websites to which they link. Those results, commonly called ‘advertisements’ or ‘ads’, are also related to the search carried out by the user and are distinguished from the natural results of general or specialised searches, for example by the words ‘ad’ or ‘sponsored’. The display of ads is linked to payment commitments from advertisers undertaken in auctions, which are managed via Google’s auction platform.

A.      Google’s online search advertising intermediation services and contracts

5        Since 2003, Google has also managed an advertising intermediation platform called AdSense. Google has developed, in that respect, various services including, among others, an online search advertising intermediation service called AdSense for Search (‘AFS’). AFS allows publishers of third-party websites independent of Google, whose websites contain integrated search engines, to display online search ads from Google when users submit queries on their websites.

6        Thus, the providers of online search advertising intermediation services (‘intermediaries’) allow website publishers to display ads linked to the online queries that users submit on websites containing an integrated search engine. In this way, intermediaries and publishers can share the revenues generated by the display of those ads.

7        As for AFS, advertisers had to associate their ads with keywords that users of the websites concerned were likely to use in a query. In order to determine which advertisers could see their ads displayed in response to an online query, Google primarily took into account, on the one hand, the price that each of those advertisers had indicated that they were willing to pay in an auction held for that purpose and, on the other hand, the relevance of those ads to the said query and, thus, the probability that the user will click on those ads. In principle, the advertiser paid the price resulting from the display of its ad only where the user actually clicked on it, such that the advertising revenues generated by that display did not depend solely on the amount of the auction at issue.

8        To use AFS, publishers could conclude inter alia two types of contracts with Google.

9        First, publishers could conclude, for one or more of their websites, an ‘online contract’, namely a standard-form, non-negotiable contract. Google classified publishers that had concluded such a contract as ‘online partners’.

10      Second, publishers could conclude, for one or more of their websites, a ‘Google Services Agreement’ (‘GSA’). Unlike online contracts, GSAs were negotiated with each publisher individually. Google classified publishers that had opted to conclude a GSA as ‘direct partners’.

11      Google drew up template GSAs, even though GSAs were contracts that were negotiated individually with direct partners. These templates were amended on a number of occasions, notably in March 2009. Moreover, in order to conclude a GSA, direct partners were required to complete an order form in which they specified whether they wished to use AFS or a different AdSense service as well as the list of website addresses for which they wished to use the requested service or services.

12      Up until March 2009, the template GSA contained inter alia two clauses. The first clause (‘the exclusivity clause’) stipulated that the direct partner was not to implement on the websites listed in the order form a service which was the same or substantially similar to the services supplied by Google under the GSA or which was in direct competition with those services. The second clause (‘the English clause’) stipulated that, subject to the exclusivity clause, the direct partner and Google had to endeavour to reach agreement on a new order form before approaching another search or advertising service provider. Moreover, in the event that the direct partner and Google did not manage to reach agreement on a new order form and the direct partner decided to approach such a provider, that clause provided that Google could make an offer matching the terms proposed by that provider.

13      From March 2009, the template GSA no longer contained either the exclusivity clause or the English clause. Instead, there were two new clauses. The first (‘the placement clause’) stipulated that, for websites using AFS, the direct partner was, first, to display a minimum number of online search ads from Google and, second, not to display such ads from other intermediaries (‘competing ads’) above those from Google or directly adjacent to them. The second (‘the prior authorisation clause’) required direct partners to obtain Google’s approval before changing the display of all online search ads, including competing ads, appearing on their results pages. It was also specified that Google could refuse to give its approval only for certain reasons and that it was assumed that it had given its approval if it did not respond within 15 business days.

14      All GSAs containing the prior authorisation clause also contained the placement clause. However, all GSAs containing the placement clause did not necessarily contain the prior authorisation clause.

15      Last, the order form corresponding to the March 2009 template GSA provided for the inclusion of screenshots of the search results pages of the websites listed in that order form for the use of AFS (‘the mock-ups’). The mock-ups had to illustrate the number, format and placement of Google search ads on those pages.

B.      Administrative procedure

16      In January 2010, the Bundeskartellamt (Federal Cartel Office, Germany) transferred to the European Commission a complaint that had been lodged by Ciao GmbH against Google.

17      On 30 November 2010, the Commission initiated proceedings against Google pursuant to Article 2(1) of Regulation (EC) No 773/2004 of 7 April 2004 relating to the conduct of proceedings by the Commission pursuant to Articles [101] and [102 TFEU] (OJ 2004 L 123, p. 18).

18      On 31 March 2011, 30 March 2012 and 30 January 2013, respectively, Microsoft Corporation, Expedia Inc. and Initiative for a Competitive Online Marketplace lodged complaints against Google.

19      On 13 March 2013, the Commission adopted a preliminary assessment, within the meaning of Article 9 of Council Regulation (EC) No 1/2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles [101] and [102 TFEU] (OJ 2003 L 1, p. 1), which related, among others, to the contractual obligations imposed by Google in respect of online search advertising campaigns.

20      On 3 April 2013, 21 October 2013 and 31 January 2014, Google offered the Commission commitments to address the concerns set out in the latter’s preliminary assessment.

21      On 16 May 2014 and 2 July 2015, respectively, Deutsche Telekom AG, and [confidential](1) and its subsidiary, [confidential], lodged complaints against Google.

22      On 21 April 2016, Microsoft and Ciao’s complaints against Google were withdrawn.

23      On 14 July 2016, the Commission decided to initiate the proceedings provided for in Article 2(1) of Regulation No 773/2004 with respect to the exclusivity, placement and prior authorisation clauses that were provided for by GSAs. On the same day, it adopted a statement of objections pursuant to Article 10 of that regulation, whereby it informed Google that those clauses were liable to constitute an abuse of a dominant position and therefore infringe Article 102 TFEU.

24      On 9 September 2016, Google informed the Commission that it had sent letters to all direct partners notifying them that it was removing the exclusivity and prior authorisation clauses in their entirety as well as certain amendments of the placement clause.

25      On 20 March 2019, the Commission adopted the contested decision.

C.      The contested decision

26      By the contested decision, the Commission found that Google had committed three separate infringements of Article 102 TFEU constituting, together, a single and continuous infringement of that provision. It accordingly ordered Google LLC to pay a fine, part of which jointly and severally with Alphabet Inc.

1.      Market definition

27      The Commission considered that the national markets for online search advertising and the EEA-wide market for online search advertising intermediation constituted distinct relevant markets for the purposes of its analysis.

(a)    National markets for online search advertising

(1)    Product market

28      The Commission considered that the provision of online search advertising, that is to say, online ads which are displayed following keyword searches which users perform on websites containing a search engine, constituted a distinct product market.

29      The Commission explained that that market required user queries to be matched with relevant search ads connected to those queries by online search advertising platforms. It also observed that, in that market, the demand was made up of users and advertisers and the supply was made up of the operators of online search advertising platforms. Those platforms necessitated, according to the Commission’s explanations, a general search service, the technology to match user queries with relevant search ads linked to those queries and an advertiser base large enough to compete against other online search advertising platforms.

30      In concluding that the online search advertising market was a distinct product market, the Commission distinguished online search ads from three other types of advertisements.

31      First, the Commission considered that offline advertising, such as ads on television, radio and in newspapers, and online advertising were not substitutable.

32      Second, the Commission considered that online search ads and online non-search ads, that is to say, ads which are placed directly on a page of a website without any connection to the keyword searches performed by users, were not substitutable.

33      Third, the Commission considered that online search ads and paid specialised search results, which involve the paid listing of advertisers’ products and services, for example on Google’s general search websites via the ‘Google Shopping’ and ‘Google Hotel Finder’ services, were not substitutable.

(2)    Geographic market

34      From a geographic perspective, the Commission considered that the online search advertising market was national, by identifying national markets within the EEA.

(b)    Market for online search advertising intermediation in the EEA

(1)    Product market

35      The Commission noted that the provision of online search advertising intermediation services, that is to say, services which, like AFS, enable publishers to ‘sell’ advertising space on their websites to advertisers wishing to display online search ads, constituted a distinct product market.

36      First, the Commission considered that there was limited substitutability between – according to the wording used in the contested decision – the ‘sale’ of online ads through an intermediary and the sale of online ads made directly by publishers.

37      Second, the Commission found that there was limited substitutability between intermediation services for online search ads and those for online non-search ads.

(2)    Geographic market

38      From a geographic perspective, the Commission considered that the online search advertising intermediation market covered the entirety of the EEA.

2.      Dominant position

39      The Commission noted that Google had held a dominant position (i) in 30 out of 31 national markets for online search advertising in the EEA in various periods between 2006 and 2016 and (ii) in the EEA-wide market for online search advertising intermediation from 2006 to 2016.

(a)    National markets for online search advertising

40      The Commission considered that Google had held a dominant position during various periods, between 2006 and 2016, on all of the national markets for online search advertising in the EEA, with the exception of Portugal, on account of its market shares, the barriers to entry and expansion and the lack of countervailing buyer power on the part of advertisers.

(1)    Market shares

41      First, the Commission calculated Google’s market shares on the basis of both its gross and its net revenues. It found that Google had held more than [confidential]% of market shares from 2006 to 2016 on all of the national markets of the EEA for which it had information, with the exception of the Czech Republic, Portugal, Slovenia, Finland, Sweden and Norway. It added that, in 2016, Google had a market share higher than [confidential]%, on the basis of its gross revenues, and higher than [confidential]%, on the basis of its net revenues, on all of the national markets of the EEA for which it had information, including the Czech Republic, Slovenia, Finland and Sweden.

42      Second, the Commission calculated Google’s market shares on the basis of the number of online queries. It found that Google had held more than [confidential]% of market shares, between 2010 and 2013, on all of the national markets of the EEA for which it had information.

43      Third, the Commission considered that, from 2006 to 2016, Google had faced limited competition from other online search advertising providers, including Microsoft and Yahoo!, despite the latter’s acquisition, in 2003, of Overture Services Inc., which was at that time the incumbent and a leader in the field.

(2)    Barriers to entry and expansion

44      The Commission considered that there were numerous barriers to entry and expansion in the national markets for online search advertising.

45      First, the Commission noted that significant investments were required in order for an online search advertising provider to be able to establish itself and that that finding applied also to online non-search advertising providers.

46      Second, the Commission considered that the national markets for online search advertising were characterised by network effects.

47      On the one hand, the Commission noted that, the greater the number of advertisers that used an online search advertising provider’s service, the more online search ads that provider could choose from and thereby increase the relevance of the ads that it served in response to a user’s query.

48      On the other hand, the Commission found that, the higher the number of users of a general search service, the greater the likelihood that an online search ad would be matched to an interested user.

49      Third, the Commission considered that the ‘strength’ of Google’s general search service and its ‘interaction’ with online search advertising could not be easily matched by competing online search advertising providers. It noted, in that regard, that Google’s general search service held, in 2016, a market share higher than [confidential]% in each of the EEA Member States, with the exception of the Czech Republic where it was still above [confidential]%.

50      Fourth, the Commission found that nearly all advertisers used Google’s auction platform, AdWords, associated with Google’s general search service.

51      Fifth, the Commission noted that, since Microsoft’s launch of adCenter in 2006, no significant entry had taken place into any national market for online search advertising in the EEA.

52      Sixth, the Commission noted that Google had strengthened its dominant position by concluding, in October 2015, an agreement with Yahoo! Inc. pursuant to which the latter would be provided with online search ads, general search services and specialised image search services.

(3)    Lack of countervailing buyer power

53      The Commission considered that the national markets for online search advertising in the EEA were characterised by a lack of countervailing buyer power on the part of advertisers.

54      First, the Commission found, on the one hand, that each advertiser represented only a small part of the total demand in the national markets for online search advertising in the EEA and, on the other hand, that advertisers could not rely solely on the advertising platforms of Google’s competitors.

55      Second, the Commission noted that advertisers could not negotiate the terms of their agreements with Google for the provision of online search advertising services and that Google imposed high prices on them.

(b)    Market for online search advertising intermediation in the EEA

56      The Commission considered that Google had held a dominant position from 2006 to 2016 on the markets for online search advertising intermediation in the EEA, in view of its market shares, the barriers to entry and expansion and the lack of countervailing buyer power on the part of publishers.

(1)    Market shares

57      First, relying on Google’s gross revenues, the Commission found, on the one hand, on the basis of data provided by Google, that it had held market shares which had always been above [confidential]% between 2006 and 2016 and which had reached, during that latter year, [confidential]% and, on the other hand, on the basis of data provided by Google, Microsoft and Yahoo!, that Google had held market shares above [confidential]% in 2006, which had always been above [confidential]% between 2007 and 2014.

58      Second, relying on Google’s net revenues, the Commission found, on the one hand, on the basis of data provided by Google, that it had held market shares above [confidential]% in 2006 and above [confidential]% between 2007 and 2016 and, on the other hand, on the basis of data provided by Google and Yahoo!, that Google had held market shares above [confidential]% between 2006 and 2011 and which had reached, during that latter year, over [confidential]%.

59      Third, the Commission inferred from Google’s market shares that it had faced limited competition from other intermediaries.

(2)    Barriers to entry and expansion

60      The Commission considered that there were numerous barriers to entry and expansion in the market for online search advertising intermediation in the EEA.

61      First, the Commission noted that significant investments were required in order to establish, maintain and refine a ‘search advertising platform’.

62      Second, the Commission considered that the online search advertising intermediation market was characterised by network effects. In that regard, it noted that the success of an intermediary depended on the number of advertisers and publishers that it could attract as well as the size of its portfolio of online search ads. It stated that those different elements were interlinked, meaning that an intermediary not attracting a sufficient number of publishers would also fail to attract sufficient advertisers. It also found that the greater the number of advertisers that used an online search advertisement intermediation service, the more ads related to those searches the intermediary could choose from and, thus, increase the relevance of the ads it served in response to a user’s query.

63      Third, the Commission noted that, since December 2009 and the partnership established by Microsoft and Yahoo!, there had been no significant entry into the EEA-wide market for online search advertising intermediation. It added that ‘a number’ of Google’s competing intermediaries had been marginalised or had exited that market since 2007.

(3)    Lack of countervailing buyer power

64      The Commission considered that the online search advertising intermediation market in the EEA was characterised by a lack of countervailing buyer power on the part of publishers.

65      First, the Commission found, on the one hand, that each publisher represented only a small part of the total demand of the EEA-wide market for online search advertising intermediation and, on the other hand, that publishers could not rely solely on the services of competing intermediaries since AFS generated the highest revenues for them.

66      Second, the Commission found that Google, on the one hand, had ceased to guarantee publishers a minimum revenue since 2013 and, on the other hand, had reduced the average revenue that it shared with publishers between 2007 and 2016.

3.      Exclusivity clause in GSAs in which direct partners had ‘typically’ included all of their websites

67      The Commission considered that the exclusivity clause had constituted, from 1 January 2006 until 31 March 2016, an abuse of a dominant position in so far as that clause was contained in GSAs in which direct partners had ‘typically’ included all of their websites displaying online search ads. Primarily, it considered that that clause imposed, in those conditions, an exclusive supply obligation on those direct partners which, as such, was contrary to Article 102 TFEU. In the alternative, it considered that the latter clause, to the extent that it imposed such an obligation, was contrary to the said provision, on the ground that it was capable of restricting competition. It categorised as ‘all sites direct partners’ direct partners that had typically included all of their websites in at least one of their GSAs.

(a)    Exclusivity clause in GSAs concluded with all sites direct partners to the extent that it constituted an exclusive supply obligation contrary, as such, to Article 102 TFEU

68      First, the Commission recalled the case-law of the Court of Justice arising from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36, paragraph 89), according to which ‘an undertaking which is in a dominant position on a market and ties purchasers – even if it does so at their request – by an obligation or promise on their part to obtain all or most of their requirements exclusively from the said undertaking abuses its dominant position within the meaning of Article [102 TFEU]’.

69      Second, the Commission found that the exclusivity clause constituted, in the present case, an exclusive supply obligation because it obliged all sites direct partners to source all or most of their requirements in terms of online search advertising intermediation services from Google. In that regard, it noted, first, that the exclusivity clause in GSAs concluded with those direct partners applied ‘typically’ to all of their websites displaying online search ads, next, that the said direct partners could not derogate from that clause before the end of their GSAs and, last, that the GSAs concluded with two of those direct partners, namely [confidential] and [confidential], required them to make all of their websites displaying such ads subject to the said clause.

70      In those conditions, the Commission found, primarily, that the exclusivity clause in GSAs concluded with all sites direct partners was contrary to Article 102 TFEU, without it having been necessary to verify whether that clause had been capable of restricting competition in the light of all the circumstances of the case.

(b)    Exclusivity clause in GSAs concluded with all sites direct partners to the extent that it constituted an exclusive supply obligation capable of restricting competition within the meaning of Article 102 TFEU

71      The Commission considered, in the alternative, that the exclusivity clause in GSAs concluded with all sites direct partners constituted an exclusive supply obligation which had been capable of restricting competition, in the light of all the circumstances of the case.

72      The Commission noted that the exclusivity clause in GSAs concluded with all sites direct partners had (i) deterred those direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers. In addition, it found that the English clause had exacerbated the capability of that clause to restrict competition.

73      The Commission added that, as part of the analysis of all the relevant circumstances, it had taken into account (i) the extent of Google’s dominant position in, on the one hand, the national markets for online search advertising in the EEA, with the exception of Portugal, and, on the other hand, the market for online search advertising intermediation in the EEA, (ii) the share of the latter market covered by the exclusivity clause in GSAs concluded with all sites direct partners, and (iii) the ‘duration of [that c]lause’.

(c)    Absence of objective justification

74      The Commission rejected the objective justification put forward by Google during the administrative procedure.

75      First, Google had claimed, during the administrative procedure, that the exclusivity clause in GSAs concluded with all sites direct partners was necessary to guarantee it a level of revenue sufficient to support, on the one hand, the investments necessary to maintain and improve its online search advertising intermediation services and, on the other hand, the specific investments made in favour of the said direct partners.

76      Second, Google had claimed that AFS had delivered procompetitive benefits by improving the quality of the user experience and by increasing advertising revenues, the usefulness of search results pages for publishers and exposure of advertisers to users interested in their products.

77      On the one hand, the Commission considered that Google had not demonstrated that the investments it cited would not have been made but for the exclusivity clause in GSAs concluded with all sites direct partners. In that regard, it noted, in essence, that the fact that Google had replaced the said clause with the placement and prior authorisation clauses confirmed that it could have made the said investments with less restrictive clauses. On the other hand, it found that the procompetitive benefits alleged by Google were not relevant for the purposes of assessing whether the exclusivity clause was objectively justified.

4.      Placement clause

78      The Commission considered that, from 31 March 2009 until 6 September 2016, the placement clause constituted an abuse of a dominant position, on the ground that that clause was capable of restricting competition in the light of all the circumstances of the case, and that Google had not demonstrated that the said clause was objectively justified.

(a)    Scope of the placement clause

79      In the first place, the Commission considered that the placement clause reserved the most prominent space on the websites of partners covered by that clause for Google search ads.

80      In the second place, the Commission considered that the placement clause required direct partners to display, in the most prominent space on their websites covered by that clause, (i) a ‘block’ of three ‘wide format’ Google search ads when the query was made on a desktop device and (ii) at least one Google search ad when the query was made on a mobile device.

(b)    Restriction of competition as a result of the placement clause

81      The Commission noted that the placement clause had (i) deterred direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers. In addition, it found that the binding nature of the mock-ups had exacerbated the capability of the said clause to restrict competition.

82      The Commission added that, as part of the analysis of all the relevant circumstances, it had taken into account (i) the extent of Google’s dominant position in, on the one hand, the national markets for online search advertising in the EEA, with the exception of Portugal, and, on the other hand, the market for online search advertising intermediation in the EEA, (ii) the share of the latter market covered by the placement clause, and (iii) the ‘duration of [that c]lause’.

(c)    Absence of objective justification

83      The Commission rejected the objective justification put forward by Google during the administrative procedure.

84      First, Google had claimed, during the administrative procedure, that the placement clause was necessary to guarantee it, to some degree, revenues sufficient to justify the investments made in favour of direct partners and to maximise their revenues (see paragraph 75 above).

85      Second, Google had claimed that a certain degree of consistency in the placement of online search ads was necessary to help maintain the relevance of those ads.

86      The Commission noted that Google had not demonstrated that the investments that it invoked would not have been made in the absence of the placement clause. In addition, it found that the circumstance, alleged by Google, that the said clause had increased the advertising revenues of direct partners was irrelevant to determining the existence of an infringement of Article 102 TFEU. Last, it considered that Google could have maintained the relevance of online search ads by using less restrictive methods, such as guidelines or content policies. It stated, in that regard, that the fact that, in 2016, Google had waived certain amendments of the clause in question (see paragraph 24 above) confirmed that it could have implemented less restrictive measures.

5.      Prior authorisation clause

87      The Commission considered that, from 31 March 2009 until 6 September 2016, the prior authorisation clause constituted an abuse of a dominant position, on the ground that that clause had been capable of restricting competition in the light of all the circumstances of the case, and that Google had not demonstrated that the said clause was objectively justified.

(a)    Restriction of competition as a result of the prior authorisation clause

88      The Commission noted that the prior authorisation clause had (i) deterred direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers.

89      The Commission added that, as part of the analysis of all the relevant circumstances, it had taken into account (i) the extent of Google’s dominant position in, on the one hand, the national markets for online search advertising in the EEA, with the exception of Portugal, and, on the other hand, the market for online search advertising intermediation in the EEA, (ii) the share of the latter market covered by the prior authorisation clause, and (iii) the ‘duration of [that c]lause’.

(b)    Absence of objective justification

90      The Commission rejected the objective justification put forward by Google during the administrative procedure.

91      Google had claimed, during the administrative procedure, that the prior authorisation clause was necessary to enable direct partners to display competing ads in compliance with its quality standards, in particular in order to avoid the display of ads posing as Google ads and promoting inappropriate content or leading to the installation of malicious software on the user’s computer.

92      The Commission considered that Google had not explained why direct partners had to display competing search ads in compliance with its quality standards or how the prior authorisation clause helped to avoid deceptive practices on the websites concerned. In addition, it was of the view that Google could have achieved compliance with its quality standards and protection of its brand using less restrictive methods. It moreover stated, in that regard, that the fact that, in 2016, Google had waived the said clause (see paragraph 24 above) confirmed that it could have implemented less restrictive measures.

6.      Single and continuous infringement

93      The Commission concluded that the three abuses of a dominant position, resulting respectively from the exclusivity clause in GSAs concluded with all sites direct partners, from the placement clause and from the prior authorisation clause, together constituted a single and continuous infringement of Article 102 TFEU which had lasted from 1 January 2006 until 6 September 2016.

94      In that regard, first, the Commission considered that the exclusivity clause in GSAs concluded with all sites direct partners, the placement clause and the prior authorisation clause pursued the same objective, namely to foreclose Google’s competing intermediaries in order to maintain and strengthen its position on the online search advertising intermediation market and the online search advertising markets and, consequently, its position on the general search market.

95      Second, the Commission considered that the exclusivity clause in GSAs concluded with all sites direct partners, the placement clause and the prior authorisation clause were complementary in that those clauses sought to deter direct partners from sourcing competing search ads and to prevent Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA. In that regard, the Commission stated, inter alia, that Google itself had referred to the placement clause as a ‘relaxed exclusivity’ clause, and that all GSAs containing the prior authorisation clause also contained the placement clause.

7.      Effect on trade between Member States

96      The Commission considered that the exclusivity clause in GSAs concluded with all sites direct partners, the placement clause and the prior authorisation clause were capable – individually and collectively – of having an appreciable effect on trade between Member States.

8.      The fine

97      The Commission ordered Google LLC to pay a fine of EUR 1 494 459 000, EUR 130 135 475 of which jointly and severally with Alphabet Inc.

II.    Forms of order sought

98      Google claims that the Court should:

–        primarily, annul the contested decision;

–        in the alternative, annul or reduce the amount of the fine;

–        order the Commission to pay the costs.

99      Surfboard Holding BV (‘Surfboard’) and Vinden.NL BV (‘Vinden’), regard being had to the observations of the latter on the report for the hearing, claim that the Court should:

–        annul the contested decision;

–        order the Commission to pay the costs.

100    The Commission contends that the Court should:

–        dismiss the action;

–        order Google to pay the costs;

–        order Surfboard and Vinden to bear the costs relating to their respective interventions.

III. Law

101    Google raises five pleas in law in support of its action, alleging (i) that the Commission did not properly define the market for online search advertising intermediation and, consequently, did not prove that Google had a dominant position on that market, (ii) that the exclusivity clause in GSAs concluded with all sites direct partners did not constitute an abuse of a dominant position, (iii) and (iv) that the placement and prior authorisation clauses did not constitute such abuses, and (v) that the Commission was wrong to impose a fine on it.

A.      Preliminary observations

102    Under Article 102 TFEU, any abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it is to be prohibited as incompatible with the internal market in so far as it may affect trade between Member States.

103    As follows from the consistent case-law of the Court of Justice, the purpose of that provision is to prevent competition from being restricted to the detriment of the public interest, individual undertakings and consumers, by sanctioning the conduct of undertakings in a dominant position that has the effect of hindering competition on the merits and is thus likely to cause direct harm to consumers, or which causes them harm indirectly by hindering or distorting that competition (judgment of 21 December 2023, European Superleague Company, C‑333/21, EU:C:2023:1011, paragraph 124; see also, to that effect, judgment of 27 March 2012, Post Danmark, C‑209/10, EU:C:2012:172, paragraph 20). The concept of ‘abuse’, within the meaning of that provision, is thus intended to penalise the conduct of a dominant undertaking which, on a market where the degree of competition is already weakened because of the presence of the undertaking concerned, adversely affects an effective competition structure (see judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 36 and the case-law cited).

104    Dominant undertakings, therefore, irrespective of the reasons for which they have such a position, have a special responsibility not to allow their behaviour to impair genuine, undistorted competition on the internal market (see judgments of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 135, and of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 38).

105    However, it is not the purpose of Article 102 TFEU to prevent an undertaking from acquiring, on its own merits, a dominant position on a market, or to ensure that competitors less efficient than an undertaking in such a position should remain on the market (judgment of 21 December 2023, European Superleague Company, C‑333/21, EU:C:2023:1011, paragraph 126; see also, to that effect, judgment of 27 March 2012, Post Danmark, C‑209/10, EU:C:2012:172, paragraph 21). Indeed, not every exclusionary effect is necessarily detrimental to competition, since competition on the merits may, by definition, lead to the departure from the market or the marginalisation of competitors that are less efficient and so less attractive to consumers from the point of view of, among other things, price, choice, quality or innovation (see judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 37 and the case-law cited; see also, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraphs 133 and 134).

106    Thus, abuse of a dominant position could be established, inter alia, where the conduct complained of produced exclusionary effects in respect of competitors that were as efficient as the perpetrator of that conduct in terms of cost structure, capacity to innovate, quality, or where that conduct was based on the use of means other than those which come within the scope of ‘normal’ competition, that is to say, competition on the merits (see judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 39 and the case-law cited).

107    In that regard, it is for the Commission to demonstrate the abusive nature of conduct in the light of all the relevant factual circumstances surrounding the conduct in question, which includes those highlighted by the evidence adduced in defence by the undertaking in a dominant position (see judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 40 and the case-law cited).

108    It is true that, in order to establish that conduct is abusive, the Commission does not necessarily have to demonstrate that that conduct actually produced anticompetitive effects. The purpose of Article 102 TFEU is to penalise abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it, irrespective of whether such practice has proved successful. Accordingly, the Commission may find that there has been an infringement of Article 102 TFEU by establishing that, during the period in which the conduct in question was implemented, that conduct had, in the circumstances of the case, the ability to restrict competition on the merits, despite its lack of effect (see judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 41 and the case-law cited).

109    However, that demonstration must, in principle, be based on tangible evidence which establishes, beyond mere hypothesis, that the practice in question is actually capable of producing such effects, since the existence of doubt in that regard must benefit the undertaking which engages in such a practice (see judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 42 and the case-law cited).

110    It should be noted that, unlike, for example, a prospective analysis of the kind required for the examination of a proposed concentration, which makes it necessary to predict events which are more or less likely to occur in future if a decision prohibiting the planned concentration or laying down the conditions for it is not adopted, it is usually a matter for the Commission, when penalising an abuse of a dominant position, to examine past events, for which there is generally a variety of evidence available that makes it possible to understand the causes and to assess the effects on effective competition (see, to that effect, judgment of 15 February 2005, Commission v Tetra Laval, C‑12/03 P, EU:C:2005:87, paragraph 42).

111    To that end, the Commission may, inter alia, by virtue of Article 18 of Regulation No 1/2003, require undertakings to provide all the information necessary for its investigation. As the Commission contends, it is necessary to attach a great evidential value to the exhaustive responses to a direct question provided under that provision in so far as undertakings which supply incorrect or misleading information in response to such a question may have a fine imposed on them under Article 23(1)(a) of that regulation (see, to that effect, judgments of 16 September 2013, Galp Energía España and Others v Commission, T‑462/07, not published, EU:T:2013:459, paragraph 123, and of 26 January 2022, Intel Corporation v Commission, T‑286/09 RENV, under appeal, EU:T:2022:19, paragraph 376).

112    In addition, it should be recalled that the undertaking concerned may submit, during the administrative procedure, on the basis of supporting evidence, that its conduct was not capable of restricting competition and, in particular, of producing the alleged foreclosure effects (see, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 138). The General Court must thus examine all of the applicant’s arguments seeking to call into question the validity of the Commission’s findings as to the foreclosure capacity of competitors that are at least as efficient, relating to the practice in question (judgment of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments), T‑235/18, EU:T:2022:358, paragraph 356, and see also, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 141).

113    With regard to the judicial review provided for in Article 263 TFEU, the Court of Justice has recalled that it extends to all the elements of Commission decisions relating to proceedings under Article 102 TFEU, which are subject to in-depth review by the General Court, in law and in fact, in the light of the pleas raised by the applicant and taking into account all the relevant elements submitted by the applicant, whether those elements pre-date or post-date the contested decision, whether they were submitted previously in the context of the administrative procedure or, for the first time, in the context of the proceedings before the General Court, in so far as those elements are relevant to the review of the legality of the Commission decision (see judgments of 21 January 2016, Galp Energía España and Others v Commission, C‑603/13 P, EU:C:2016:38, paragraph 72, and of 25 July 2018, Orange Polska v Commission, C‑123/16 P, EU:C:2018:590, paragraph 105 and the case-law cited).

B.      First plea: the Commission erroneously defined the relevant markets at issue and Google’s dominant position

114    By the first plea, Google, supported by Surfboard and Vinden, claims that the Commission made an erroneous definition of the markets for online search advertising and online search advertising intermediation. That error means that the Commission did not prove that Google had a dominant position on the market for online search advertising intermediation in the EEA, such that it could not validly conclude, in the contested decision, that there was an abuse within the meaning of Article 102 TFEU on that market.

115    Google’s line of argument is divided into two parts.

116    In the first part of its first plea, Google claims that, for the purposes of defining the national markets for online search advertising, the Commission erroneously concluded that online non-search ads and online search ads were not substitutable.

117    In the second part of its first plea, Google claims that, for the purposes of defining the European market for online search advertising intermediation, the Commission erroneously concluded that there was no substitutability between online ads sold via an intermediary and online ads sold directly by publishers.

1.      First part of the first plea: substitutability between online search ads and online non-search ads

118    As has been noted in paragraph 27 above, the Commission defined two relevant markets in the contested decision. The first of those markets was the market for online search advertising, described in paragraphs 28 and 29 above. Although the abuse of a dominant position found in the contested decision took place on the second relevant market defined in that decision, namely the market for online search advertising intermediation, the Commission explained during the hearing that defining the first market was a necessary step for defining the second, which cannot be defined as a distinct market without previously having defined the first.

119    In the contested decision, the Commission inter alia considered that online search ads and online non-search ads, described in paragraph 32 above, were not substitutable.

120    The Commission based its analysis distinguishing online search ads from online non-search ads (‘the two types of ad at issue’), in recitals 135 to 169 of the contested decision, on (i) the triggering and positioning of the ads in question, (ii) their formats, (iii) their capabilities to answer to an immediate interest of the user, (iv) their capabilities to induce the user to make a purchase, (v) their click-through rates (‘CTRs’) and conversion rates, (vi) their capabilities to measure the return on investment of advertisers, (vii) the observations of an association representing advertisers (World Federation of Advertisers), (viii) the replies of advertisers, publishers and media agencies to the Commission’s request for information concerning the effect of an increase in the price of online search ads, (ix) an industry report by an online database operator providing statistical data and research (Statista), and (x) the investments necessary for the provision of online search advertising services.

121    Google considers that the Commission erroneously concluded, in the contested decision, that the two types of ad at issue were not in the same market.

122    Google’s line of argument in that regard may be summarised as follows. First, it notes that the Commission focused – wrongly – on the substitutability of the two types of ad at issue from an advertiser’s perspective rather than from that of a publisher. Second, the Commission did not take into account all relevant factors, focusing on alleged differences in characteristics between the two types of ad at issue. Third, the Commission did not carry out an adequate price analysis, for example by means of a test analysing the impact of a significant and non-transitory increase of 5 to 10% in the price of online search ads (‘the SSNIP test’), and misinterpreted the replies of publishers, advertisers and media agencies in the analysis it carried out. Fourth, those alleged differences in characteristics between the two types of ad at issue are not borne out and, moreover, do not suffice to conclude that they were not substitutable. Fifth, the Commission did not take into account examples of publishers which switch or switched between the two types of ad at issue. Sixth, the Commission misinterpreted the statements of certain Google representatives. Seventh, the analysis conducted in the contested decision of the substitutability of the two types of ad at issue is, in essence, contrary to past Commission decisions.

123    Surfboard and Vinden support Google’s arguments and also claim that the Commission did not properly take the perspective of publishers into account.

(a)    Preliminary observations

124    As a preliminary point, it is appropriate to recall that the definition of the relevant market, in the application of Article 102 TFEU, is, as a general rule, a prerequisite of any assessment of whether the undertaking concerned holds a dominant position, the objective being to define the boundaries within which it must be assessed whether that undertaking is able to behave, to an appreciable extent, independently of its competitors, customers and consumers (see judgment of 30 January 2020, Generics (UK) and Others, C‑307/18, EU:C:2020:52, paragraph 127 and the case-law cited).

125    It is clear from the case-law that the concept of the relevant market implies that there can be effective competition between the products or services which form part of it and this presupposes that there is a sufficient degree of interchangeability of all the products or all the services forming part of the same market in so far as a specific use of such products or services is concerned. Interchangeability or substitutability is not assessed solely in relation to the objective characteristics of the products and services at issue. The competitive conditions and the structure of supply and demand on the market must also be taken into consideration (see judgment of 23 January 2018, F. Hoffmann-La Roche and Others, C‑179/16, EU:C:2018:25, paragraph 51 and the case-law cited).

126    It also follows from the Commission Notice on the definition of relevant market for the purposes of Community competition law (OJ 1997 C 372, p. 5; ‘the market definition notice’), that ‘a relevant product market comprises all those products and/or services which are regarded as interchangeable or substitutable by the consumer, by reason of the products’ characteristics, their prices and their intended use’. From an economic point of view, for the definition of the relevant market, demand substitution constitutes the most immediate and effective disciplinary force on the suppliers of a given product, in particular in relation to their pricing decisions. Supply-side substitutability may also be taken into account when defining markets in those situations in which its effects are equivalent to those of demand-side substitution in terms of effectiveness and immediacy. That means that suppliers are able to switch production to the relevant products and market them in the short term without incurring significant additional costs or risks in response to small and permanent changes in relative prices (judgment of 29 March 2012, Telefónica and Telefónica de España v Commission, T‑336/07, EU:T:2012:172, paragraph 113).

127    Furthermore, it should be emphasised that, as is apparent from paragraph 25 of the market definition notice and from the case-law, the definition of the relevant market does not require the Commission to follow a rigid hierarchy of different sources of information or types of evidence (judgment of 11 January 2017, Topps Europe v Commission, T‑699/14, not published, EU:T:2017:2, paragraph 82), the question whether products are substitutable being liable to be determined on the basis of a range of evidence consisting of various items, often of an empirical nature, and the Commission having to take into account all relevant available information (judgment of 1 July 2010, AstraZeneca v Commission, T‑321/05, EU:T:2010:266, paragraph 85).

(b)    Taking into account of the perspective of publishers

128    Google submits that the Commission failed to demonstrate, in the contested decision, that the two types of ad at issue were in different markets from a publisher perspective. According to Google, the Commission should have further examined substitutability from the side of publisher demand, since the abuse alleged by the Commission in the contested decision concerns the restriction of the possibility for publishers to choose alternatives to Google’s intermediation service, namely AFS. The Commission therefore focused wrongly in the contested decision on substitutability from the perspective of advertisers and not from that of publishers.

129    Surfboard and Vinden support those arguments.

130    The Commission disputes Google’s line of argument.

131    In the first place, it should be pointed out that substitutability must generally be examined from the demand side (see paragraph 126 above).

132    Google acknowledged during the hearing, however, that publishers did not constitute the demand for the two types of ad at issue. Moreover, Google has not disputed the Commission’s assertion, in recital 121 of the contested decision, according to which, in the market for online search advertising, demand was made up of users and advertisers and supply was made up of the operators of advertising platforms. It follows that Google has not demonstrated that the perspective of publishers – which are not on the demand side – should be taken into account in the analysis of demand-side substitutability.

133    In those conditions, Google, Surfboard and Vinden cannot criticise the Commission for having included, in its analysis of the definition of the relevant market, more elements addressing the substitutability of the two types of ad at issue from the perspective of advertisers than from that of publishers simply due to the fact that the abuse found in contested decision restricted the choice of publishers on the online search advertising intermediation market, which is, moreover, a distinct market.

134    In the second place and in any event, it is important to point out that the Commission did take into account in the contested decision the perspective of publishers during its analysis of the substitutability of the two types of ad at issue.

135    Thus, the Commission relied on the replies of publishers to its requests for information in its assessment of the definition of the relevant market in order to support its conclusions on the differences in characteristics and use between the two types of ad at issue. Such a consideration is apparent from the Commission’s observations on the positioning and format of online search ads (see recitals 136 and 137 of the contested decision), the intrinsic capability of those ads to answer to the user’s immediate interest (see recitals 138 and 139 of the contested decision) and their better suitability for converting existing demand into a purchase (see inter alia recitals 142 and 143 of the contested decision). That finding shows that the analysis of the characteristics and uses of the said ads concerned both publishers and advertisers. In addition, the Commission referred to the observations of publishers to conclude that they would be unlikely to replace all or part of their online search ads with online non-search ads on their websites in the event of a 5-10% reduction in revenues from online search ads (see, in this regard, recital 148 of the contested decision). The opinions and the conduct of publishers displaying online search ads on their websites were also referred to in recitals 156 to 158, 160, 162 and 164 of the contested decision.

136    Accordingly, Google does not demonstrate that the Commission examined in an insufficient manner the substitutability of the two types of ad at issue from the perspective of publishers.

(c)    Taking into account of all relevant factors

137    Google claims that the Commission focused – wrongly – on the alleged differences in characteristics between the two types of ad at issue, by failing to take into account all the factors relevant for publishers, as is required by the market definition notice. Google criticises in particular the Commission, relying on paragraphs 38 to 43 of that notice, for not having used, in the contested decision, actual examples of product substitution, quantitative tests to measure price elasticity, reasoned views of customers or competitors and the barriers and costs for publishers associated with switching products.

138    The Commission disputes Google’s line of argument.

139    As is apparent from the case-law cited in paragraphs 125 and 126 above, substitutability is not assessed solely in relation to the objective characteristics of the products and services at issue. While the characteristics of the products and services at issue are relevant to that assessment, the competitive conditions and the structure of supply and demand on the market must also be taken into consideration (see judgment of 23 January 2018, F. Hoffmann-La Roche and Others, C‑179/16, EU:C:2018:25, paragraph 51 and the case-law cited). The market definition notice, to which Google refers, essentially reiterates that principle in paragraph 36 thereof, indicating that, ‘however, product characteristics and intended use are insufficient to show whether two products are demand substitutes’.

140    Nevertheless, contrary to what Google seems to imply in its submissions, the Commission is not obliged to examine, when it is called upon to define a product market, all the elements of assessment set out in the market definition notice, or to follow a rigid hierarchy of evidence, as is apparent from paragraph 25 of the said notice itself and from the case-law cited in paragraph 127 above.

141    In any event, while the Commission did devote, in the contested decision, much of its analysis to the characteristics and uses that the two types of ad at issue might have and the differences between them, it must be pointed out that, in concluding that those ads are not substitutable, it did not merely mention those elements and that it took into account, in an overall assessment, a series of other factors, including those noted by Google and set out in paragraph 137 above.

142    Thus, the Commission also examined in the contested decision factors such as the price of the ads at issue (recitals 148 and 149), the investments necessary for the provision of online search advertising services (recitals 150 to 154) and, in its response to the arguments raised by Google during the administrative procedure, the conduct of publishers that had decreased their use of online search ads (recitals 162, 164 and 165) as well as Google representatives’ perception of the market (recitals 156 and 169). In addition, as the Commission contends in its written submissions, the issue of obstacles for publishers and advertisers to replacing online search ads with online non-search ads was also addressed, implicitly, in the contested decision. In recital 148 thereof, the Commission observed that all publishers and a majority of advertisers had indicated in reply to the Commission’s requests for information that they would be unlikely to replace online search ads with online non-search ads in the event of either a non-transitory 5-10% reduction in the revenues from online search ads, as far as publishers are concerned, or an equivalent increase in the price of those ads, as far as advertisers are concerned.

143    Moreover, contrary to what Google seems to argue, the Commission did collect the reasoned views of market participants, particularly following the requests for information sent to publishers, advertisers and media agencies, which run advertising campaigns for undertakings. The Commission relied on those views to define the relevant market, as is apparent from footnotes 105, 109, 110, 112 to 115, 119, 120, 122 to 125, 128, 132 to 138, 140, 141, 145, 169, 171, 172 and 176 of the contested decision.

144    It follows that Google does not demonstrate that the Commission disregarded certain relevant factors in its overall assessment of the substitutability of the two types of ad at issue, or that it committed an error of law by devoting a large part of its analysis to the differences in characteristics and uses between those ads.

(d)    Google’s arguments relating to the SSNIP test

145    Google claims that it was particularly important for the Commission to consider whether publishers and advertisers would have opted to use online non-search ads if there was a material change in the price of online search ads, by means of, for example, a SSNIP test. In such a case, the two types of ad at issue would put competitive pressure on each other, such that they would form a single market. According to Google, the price analysis carried out by the Commission, described in recital 148 of the contested decision, does not constitute a proper SSNIP test and, moreover, the Commission drew erroneous conclusions from that analysis.

146    It is appropriate to examine, in the first place, Google’s arguments disputing the adequacy of the price analysis that the Commission carried out and, in the second place, Google’s arguments addressing the conclusions that the Commission drew from that analysis.

(1)    Adequacy of the price analysis carried out by the Commission

147    In recital 148 of the contested decision, first, the Commission noted that ‘a majority of advertisers, all publishers and half of the media agencies indicate that they would be unlikely to replace all or part of their online search ads by non-search ads in the event of a non-transitory 5-10% increase in the price of online search ads’. Second, the Commission added, in the same recital, that certain publishers had also indicated that that was the case because the revenues from online search ads were far higher than those from online non-search ads.

148    Google argues that the Commission did not carry out a proper SSNIP test. In that regard, first, Google claims that the Commission did not investigate whether it would be profitable for a hypothetical monopolist to increase the price of online search ads by 5-10%. In other words, the Commission should have examined whether enough marginal customers would sufficiently switch their demand to render a price increase unprofitable. Second, it asserts that the Commission could not rely exclusively on the replies of publishers, advertisers and media agencies to a question that had been put to them in the requests for information, particularly as those replies were not backed up by factual evidence.

149    The Commission disputes Google’s line of argument.

150    As a preliminary point, it is necessary to note that it is apparent, in essence, from recital 148 of the contested decision that the Commission asked advertisers, publishers and media agencies how they would react in the event of an increase in the price of online search ads or, in the case of publishers, a reduction in revenues from those ads. That seems to be confirmed by footnotes 135 to 138, which set out the replies of certain undertakings to Question 2.2 of the request for information relating to AdWords of 22 December 2010 and to Question 2 of the request for information of the Commission of 26 July 2013 addressed to publishers, as well as to Question 12 of the request for information of 11 January 2016 addressed to publishers and to Question 9 of the request for information of the same date addressed to media agencies.

151    In that regard, it is appropriate to examine, first of all, the content of those questions contained in the Commission’s requests for information (together, ‘the question on prices’).

152    Thus, with regard to the advertisers, it is apparent from Question 2.2 of the request for information relating to AdWords of 22 December 2010 and Question 12 of the request for information of 11 January 2016 addressed to advertisers (certain replies to which are provided in Annexes B.3 and B.4 to the defence) that the Commission requested them to explain, in the event that they used online search ads, whether they would replace part or all of those ads with online non-search ads if the price of online search ads increased by 5-10% – by effect of the price mechanism and not of a difference in conversion rate – the price of online non-search advertising remaining stable, and to specify whether they would take into account factors other than price in that decision.

153    With regard to the media agencies, it is apparent from Question 9 of the request for information of 11 January 2016 addressed to media agencies (certain replies to which are provided in Annexes B.3 and B.4 to the defence) that the Commission requested them to explain, in the event that they placed online search ads for their customers, whether they would replace part or all of those ads with online non-search ads if the price of online search ads increased by 5-10% – by effect of the price mechanism and not of a difference in conversion rate – the price of online non-search advertising remaining stable, and to specify whether they would take into account factors other than price in that decision.

154    With regard to the publishers, it is apparent from the second sentence of Question 2 of the request for information of 26 July 2013 addressed to publishers (certain replies to which are presented in Annex B.6 to the defence) that the Commission requested them to explain whether they would replace part or all of the online search ads displayed on their websites with online non-search ads if the revenues generated by the display of online search ads decreased by 5-10%, the revenues generated by the display of online non-search ads remaining stable.

155    In the first place, regarding Google’s arguments in that respect, it should be recalled that, according to paragraph 17 of the market definition notice, a SSNIP test consists in examining whether the parties’ customers would switch to readily available substitutes or to suppliers located elsewhere in response to a hypothetical small (in the range 5-10%) but permanent relative price increase in the products and areas being considered. If substitution were enough to make the price increase unprofitable because of the resulting loss of sales, additional substitutes and areas are included in the relevant market.

156    In the case at hand, the Court finds that, in recital 148 of the contested decision, the Commission did not carry out a SSNIP test within the meaning of paragraph 17 of the market definition notice in that it did not examine the question whether it would be profitable for an undertaking to increase the price of online search ads by 5-10%. Instead, it carried out the price analysis described in paragraphs 150 to 154 above.

157    In that regard, it should be borne in mind that the Commission is not obliged to carry out systematically a SSNIP test when it defines the market in a decision applying the rules of competition law, even though that test constitutes, in the market definition notice, a tool recognised for that purpose.

158    Indeed, the Court has held that the Commission was not required to apply the SSNIP test, finding that, although that type of economic test is indeed a recognised method for defining the relevant market, it is not the only method available to the Commission. In that regard, it considered that the Commission could also take into account other tools for the purposes of defining the relevant market, such as market studies or an assessment of consumers’ and other competitors’ points of view (judgment of 11 January 2017, Topps Europe v Commission, T‑699/14, not published, EU:T:2017:2, paragraph 82).

159    That principle is reflected in paragraph 15 of the market definition notice, which states that the carrying out of a SSNIP test is only ‘one way’ of assessing product substitutability. Likewise, paragraph 25 of that notice indicates that ‘there is a range of evidence permitting an assessment of the extent to which substitution would take place’ and that, in that respect, ‘the Commission follows an open approach to empirical evidence, aimed at making an effective use of all available information which may be relevant in individual cases … [and] does not follow a rigid hierarchy of different sources of information or types of evidence’.

160    Furthermore, according to the case-law, the SSNIP test may also prove unsuitable in certain cases, for example in the presence of the ‘cellophane fallacy’, that is, the situation where the undertaking concerned already holds a virtual monopoly and the market prices are already at a supra-competitive level, or where there are free goods or goods the cost of which is not borne by those determining the demand (judgment of 11 January 2017, Topps Europe v Commission, T‑699/14, not published, EU:T:2017:2, paragraph 82).

161    It follows that, while it is true that the SSNIP is a recognised tool whose results may be taken into account, with other elements, in an overall market definition assessment, systematic recourse to such a test is not obligatory for the purposes of defining the market.

162    It is therefore appropriate to reject Google’s argument according to which, in essence, the price analysis carried out by the Commission in recital 148 of the contested decision was not adequate on the ground that it does not constitute a ‘proper’ SSNIP test.

163    In the second place, it must be ascertained whether the price analysis conducted by the Commission nevertheless constitutes an adequate means of defining the relevant market, as that institution maintains.

164    In that regard, it should be noted that, according to the Commission, the replies of advertisers, media agencies and publishers to the question on prices did not indicate that the mere increase by 5-10% of the price of online search ads for advertisers and media agencies or the mere decrease for publishers by 5-10% of the revenues from such ads would probably lead those operators to replace online search ads – either completely or even partly – with online non-search ads. The Commission has explained in its written submissions that the replies that it received to the question on prices revealed that such a change in prices or revenues would not in itself be a decisive factor in the choice of those operators between the two types of ad at issue. According to the Commission, factors other than price were regarded by the undertakings that had replied as equally important, if not more so, in the choice of ad to use, such as CTR, conversion rate, overall return on investment of advertising campaigns or relevance of the ad to the user.

165    Assuming, however, that the Commission’s interpretation of the content of the replies to the question on prices is founded – which will be examined in paragraph 168 et seq. below – it is appropriate to note, as the Commission does and contrary to what Google claims, that the price analysis that it carried out was a useful way of understanding how advertisers, media agencies and publishers would react in the event of an increase in the price of online search ads or, in the case of publishers, a decrease in the revenues from those ads, and thus to evaluate whether the two types of ad at issue could potentially be perceived by those actors as substitutable. Such replies from market participants, accompanied by the reasons underpinning their replies, as has been the case here, are among the elements expressly considered relevant to the definition of the market, in that they make it possible to evaluate customers’ and competitors’ perspectives and therefore constitute, in accordance with the case-law cited in paragraph 158 above as well as in paragraph 40 of the market definition notice, a tool that may be taken into account for the purpose of defining the relevant market.

166    Such learnings may thus, in principle, be useful for the assessment of the substitutability of the two types of ad at issue from the perspective of advertisers, media agencies and publishers, in particular where, as is the case here, they reflect the perspective of a large number of those advertisers, publishers and media agencies. They may thus constitute one indicator among others, in an overall market definition assessment, of the lack of substitutability of those ads, in accordance with the Commission’s arguments.

167    It is therefore appropriate to examine, in the light of Google’s arguments, the merits of the conclusions drawn by the Commission from its price analysis.

(2)    Merits of the conclusions drawn by the Commission from the price analysis that it carried out

168    In the first place, Google submits that the conclusions that the Commission drew, in recital 148 of the contested decision, from its price analysis are based on a misinterpretation of the undertakings’ responses to the question on prices. According to it, the Commission misinterpreted certain of the replies mentioned in the footnotes inserted into that recital and disregarded the replies of other undertakings that had given an opinion contrary to the conclusions of the Commission set out in the said recital and which are not mentioned in the contested decision.

169    In the second place, Google claims that the conclusions of the Commission at issue are, in essence, based on a misrepresentation of the undertakings’ replies to the question on prices.

170    The Commission disputes Google’s line of argument.

171    In that regard, it should be recalled that the Commission asserted, in recital 148 of the contested decision, that a ‘majority’ of advertisers, ‘all’ publishers and ‘half’ of the media agencies had indicated that they would be unlikely to replace all or part of their online search ads with online non-search ads in the event of a non-transitory 5-10% increase in the price of online search ads.

172    To support that finding with regard to publishers, the Commission mentioned, in footnote 136 to the contested decision, the replies of six publishers to Question 2 of the requests for information of 26 July 2013. As far as advertisers are concerned, in footnote 135 to the contested decision, the replies of 5 advertisers to Question 12 of the request for information of 11 January 2016 are set out, as are those of 10 others to Question 2.2 of the request for information of 22 December 2010 relating to AdWords. As far as media agencies are concerned, in footnote 137 to the said decision, the replies of four media agencies to Question 9 of the request for information of 11 January 2016 are mentioned, as are those of six others to Question 2.2 of the request for information of 22 December 2010 relating to AdWords.

173    In the first place, it is appropriate to examine Google’s arguments regarding the Commission’s supposed misinterpretation of the replies provided by the undertakings by assessing its arguments in relation to (i) publishers, (ii) advertisers, and (iii) media agencies. In the second place, the Court will examine Google’s arguments on the alleged misrepresentation of those replies in recital 148 of the contested decision.

(i)    Interpretation of the replies to the question on prices

–       Interpretation of the replies of publishers

174    In the first place, it must be pointed out that, of the replies mentioned in footnote 136 to the contested decision in support of the Commission’s assertion in recital 148 of that decision, Google challenges in its written submissions only the interpretation of the reply of [confidential] (‘[confidential]’). Google therefore does not criticise the interpretation of the five other replies of the publishers that are listed therein, namely [confidential], [confidential], [confidential], the [confidential] group (to which belong the websites of [confidential]) and [confidential], or the reliability of those replies, merely observing that two of them were from complainants in the case. That fact alone, however, is in no way relevant in demonstrating that the Commission misinterpreted those replies. Google does not explain why the two replies at issue are less reliable or less credible simply because they originate from complainants, in particular in the light of the case-law cited in paragraph 111 above.

175    Regarding the reply of [confidential], the complete version of which is set out in exhibit 4 of Annex B.6 to the defence, it is apparent from that reply that the publisher in question explained that the two types of ad at issue were different in several respects, that it displayed only online search ads on its websites, since they were of much greater value to it (its business consisting in operating a search engine) and that it did not know whether a ‘more typical’ publisher would shift advertising space to online non-search ads in the event of a 5-10% decrease in revenues from online search ads. In view of the content of that reply, the Commission could reasonably infer that the undertaking in question would probably not change type of ad in the event of a 5-10% decrease in revenues from online search ads.

176    In the second place, regarding Google’s argument that the Commission disregarded the replies given by [confidential] and by [confidential], it must be held that those replies likewise do not undermine the Commission’s conclusion in recital 148 of the contested decision and recalled in paragraph 171 above, so far as concerns how publishers would react in the event of a 5-10% decrease in revenues from online search ads.

177    The observations of [confidential] and of [confidential], extracts of which are presented in Annex A.13 and in exhibit 24 of Annex A.12 to the application, respectively, were made in reply to the requests for information of 22 December 2010 relating to AdWords. Those operators thus did not reply to the request for information of 26 July 2013 addressed to publishers, contrary to all of the replies mentioned in footnote 136 to the contested decision. Those operators thus replied to a question on how they would react in the event of an increase in the price of online search ads, which would affect advertisers, and not in the event of a decrease in revenues, which would affect publishers. In addition, in respect of [confidential], it must be noted that, even though that undertaking also included in its reply specific developments concerning a subsidiary, which concluded a GSA with Google as a publisher, according to recitals 348 and 355 of the contested decision, that reply still gives an advertiser’s point of view and, moreover, indicates that a price increase of 5-10% would not necessarily lead to advertising budget being transferred to online non-search ads.

178    It follows from the foregoing that Google does not present evidence capable of demonstrating that the Commission misinterpreted, in recital 148 of the contested decision, the replies of publishers to the question on prices.

–       Interpretation of the replies of advertisers

179    In the first place, it is necessary to point out that Google disputes only the interpretation of 4 of the 15 replies mentioned by the Commission in footnote 135 to the contested decision to support its assertion in recital 148 of that decision, namely the replies of [confidential], of ‘[confidential]’ and of [confidential], as well as of [confidential]. It thus criticises neither the Commission’s interpretation of the replies of the 11 other undertakings that are listed therein, namely [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential] and [confidential], nor the reliability of those replies.

180    Concerning the four replies disputed by Google, first, it must be pointed out that, although [confidential] and ‘[confidential]’ indicated that they might consider transferring part of their budget to online non-search ads in the event of a 5-10% increase in the price of online search ads, those advertisers did not indicate that such a transfer would be likely. [confidential] after all explained in its reply, a full version of which is to be found in exhibit 2 of Annex B.3 to the defence, that the two types of ad at issue were not ‘completely substitutable’ and that it would evaluate the return on investment of online search ads before ending investments for that latter type of ad. ‘[confidential]’, for its part, recalled at the beginning of its reply to the question on prices, presented in exhibit 3 of Annex A.12 to the application, that online non-search ads did not serve the consumer’s interest in the same way as online search ads, which suggests that such a shift by that undertaking would not be likely in the event of a price increase. Second, although [confidential] indicated in its reply to the question on prices, presented in exhibit 4 of Annex A.12 to the application, that it would probably lower its investment in online search ads, it also stated that it would not ‘necessarily’ increase its use of online non-search ads however, which implies that such a shift by that undertaking would not be likely, either. Third, [confidential] clearly expressed in its reply, a full version of which is to be found in exhibit 1 of Annex B.3 to the defence, the opinion that, even though it re-evaluated its strategy each time that a price change occurred, a 5-10% increase in the price of online search ads would not be sufficiently large for it to change the type of ad used.

181    It follows that those replies do not undermine the Commission’s conclusion that those undertakings indicated that they would be unlikely to replace all or part of their online search ads with online non-search ads in the event of a non-transitory 5-10% increase in the price of online search ads, as none of those advertisers stated that such a switch would be likely.

182    In the second place, Google notes that a certain number of publishers that had replied to the question on prices, the replies of which are not mentioned in the contested decision, gave an opinion contrary to the conclusion of the Commission in recital 148 of the contested decision and that the Commission simply ignored those replies.

183    In that regard, it should be recalled that the Commission found, in recital 148 of the contested decision, that a ‘majority’ of advertisers had indicated that they would be unlikely to replace all or part of their online search ads with online non-search ads in the event of a non-transitory 5-10% increase in the price of the former. This means that the Commission itself implicitly acknowledged that certain replies had not fully ruled out the possibility of a switch in the event of a price increase.

184    Furthermore, it must be pointed out that the replies cited by Google in Table 4 of Annex A.12 to the application, which concern both advertisers and media agencies, aimed at demonstrating that the opinions of operators were split in the analysis conducted by the Commission, do not support the finding that the operators behind those replies would probably replace all or part of their online search ads with online non-search ads in the event of a 5-10% increase in the price of the former type of ad.

185    First, the content of the undertakings’ replies cited by Google reveals that none of those undertakings clearly indicated that it would replace online search ads with online non-search ads on the sole ground of a 5-10% increase in the price of online search ads.

186    Second, while the majority of the undertakings mentioned indicated – in some cases only implicitly – that, in such a case of price increase, they would assess whether it was appropriate to change the type of ad used, those undertakings nevertheless clearly revealed that that decision would depend on factors other than price, such as ad performance, the objectives of the advertising campaigns in question, CTR and return on investment. That finding is illustrated by the replies of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential] and of [confidential], extracts of which are presented in Annex A.12 to the application.

187    Third, while [confidential] and [confidential], the full versions of whose replies are presented in exhibit 6 of Annex B.3 and in exhibit 1 of Annex B.4 to the defence, respectively, indicated that they would probably change type of ad, both of those undertakings also qualified their comments, explaining that such a possible change would also depend on other factors. It should be noted, moreover, that [confidential] stated that, given that it had not been using online search ads at the time of its reply, an increase in the price of those ads would not have any impact on its advertising expenditure, which tends to limit the relevance of its reply.

188    Fourth, although [confidential] and [confidential], extracts of whose replies are presented in exhibits 24 and 27 of Annex A.12 to the application, respectively, highlighted the interchangeability of the two types of ad at issue, they did not explicitly answer the question of the effect that an increase in the price of online search ads would have. In addition, [confidential] indicated that the price of online search ads had no impact on their substitutability with online non-search ads.

189    It follows from the foregoing that the responses cited by Google tend to confirm that the increase in the price of online search ads would probably not, in itself, lead advertisers to replace all or part of those ads with online non-search ads. That evidence therefore cannot counterbalance the content of the advertisers’ responses mentioned by the Commission in footnote 135 to the contested decision.

190    Accordingly, Google does not present evidence capable of demonstrating that the Commission misinterpreted, in recital 148 of the contested decision, the responses of advertisers to the question on prices.

–       Interpretation of the replies of media agencies

191    As regards the interpretation of the replies of media agencies, in the first place, Google criticises the Commission’s interpretation of 7 of the 10 replies mentioned in footnote 137 to the contested decision in support of its assertion in recital 148, namely the replies of [confidential], of [confidential], of [confidential], of [confidential], of [confidential], of [confidential] and of ‘[confidential]’. It therefore disputes neither the interpretation of the replies of the three other media agencies that are listed there, namely [confidential], [confidential] and [confidential], nor the reliability of those responses.

192    As regards the seven replies the interpretation of which is disputed by Google, first, it is apparent from five of those seven replies that the media agencies concerned did not categorically rule out the possibility of switching to online non-search ads in the event of a 5-10% increase in the price of online search ads. On the one hand, though answering in the negative the question whether they would make such a switch, [confidential] and [confidential] also suggested in their replies to the question on prices, presented in exhibits 8 and 10 of Annex A.12 to the application, that it would be possible for them to shift a limited part of their online search ads to other types of online advertising. On the other hand, [confidential], ‘[confidential]’ and [confidential] indicated in their replies to the question on prices, presented in exhibits 6, 7 and 10 of Annex A.12 to the application, respectively, that a substitution would be possible if other factors – such as return on investment – incentivised them to do that, but their replies did not suggest that such a switch would be likely in practice.

193    That being so, while those five replies are qualified in terms of how the undertakings might react to an increase in the price of online search ads, the fact remains that those undertakings did not indicate that a 5-10% price increase would probably, in itself, lead them to replace part or all of those ads with online non-search ads.

194    Second, with regard to the replies of [confidential] and of [confidential], although Google notes that [confidential] copied and pasted another reply given as part of the request for information (exhibit 9 of Annex A.12 to the application) and that [confidential] explained that it did not use online search ads (exhibit 11 of Annex A.12 to the application), the fact remains that both of those media agencies nevertheless answered in the negative the question of whether they would replace online search ads with online non-search ads in the event of a 5-10% increase in the price of the former type of ad. That being so, the Commission could correctly use those replies in support of its conclusion on the media agencies in recital 148 of the contested decision.

195    In the second place, Google notes that a number of media agencies that had replied to the question on prices gave an opinion contrary to the conclusion of the Commission in recital 148 of the contested decision and that the Commission simply ignored those replies.

196    In that regard, it should be recalled that the Commission found that half of the media agencies had indicated that they would be unlikely to replace all or part of their online search ads with online non-search ads in the event of a non-transitory 5-10% increase in the price of the former. In a similar manner to what has been observed in paragraph 183 above in regard to advertisers, the Commission therefore implicitly acknowledged that the other half of the media agencies had not supported that assertion and thus the opinions of the media agencies were split.

197    Regarding the interpretation of the replies cited by Google which the Commission allegedly ignored, it is appropriate to refer to the analysis made in paragraphs 184 to 189 above, which concerns both advertisers and media agencies. As is explained in the abovementioned paragraphs, the replies noted by Google tend to indicate that such a price increase would not, in itself, lead advertisers and media agencies to switch from online search ads to online non-search ads. Thus, those replies cannot undermine the Commission’s conclusion that half of the media agencies would be unlikely to replace all or part of their online search ads with online non-search ads in the event of a 5-10% increase in the price of the former.

198    It follows from the foregoing that Google does not present evidence capable of demonstrating that the Commission misinterpreted, in recital 148 of the contested decision, the replies of media agencies to the question on prices.

–       Conclusion on the Commission’s interpretation of the replies to the question on prices

199    It is apparent from the foregoing analysis of the replies of the publishers, advertisers and media agencies to the question on prices, and without it being necessary to rule on the admissibility, disputed by the Commission, of Google’s line of argument contained in Annexes A.12 to the application and C.3 to the reply, that Google has not demonstrated that the Commission misinterpreted those replies. Its line of argument cannot therefore call into question the conclusions that the Commission drew from them in recital 148 of the contested decision.

(ii) Alleged misrepresentation of the replies to the question on prices

200    Google also claims that the Commission incorrectly described, in recital 148 of the contested decision, the content of the replies of the undertakings to the question on prices in that its conclusions are based on the replies of only 15 advertisers, 6 publishers and 10 media agencies, which are set out in footnotes 135 to 137 to that decision. The Commission’s conclusions are therefore, in essence, flawed or misleading as they are based on replies from fewer than 10% of the advertisers mentioned in the contested decision and from fewer than 20% of the advertisers and from a third of the media agencies that had received requests for information during the administrative procedure.

201    The Commission disputes Google’s line of argument.

202    In the first place, it should be noted that the Commission has explained, in its written submissions, that the replies set out in footnotes 135 to 137, in order to support the conclusions in recital 148 of the contested decision, was not an exhaustive list of the undertakings that had provided replies. In addition to the replies of the undertakings mentioned in those footnotes, it notes (i) that 6 other publishers replied to the question on prices in the request for information of 26 July 2013, (ii) that 7 other publishers replied to the question on prices in a different request for information, that is, the one dated 18 March 2016, (iii) that 43 other advertisers answered in the negative to the question, contained in the requests for information of 22 December 2010 and 11 January 2016, of whether they would replace part of or all online search ads with online non-search ads in the event of a non-transitory 5-10% increase in the price of online search ads, and (iv) that 7 other media agencies also answered in that way. Moreover, the Commission has stated in its written submissions that it had received replies to the question on prices from a total of 19 publishers, 87 advertisers and 34 media agencies.

203    That information is absent from the contested decision, however. The Commission did not in fact specify in that decision the non-exhaustive nature of the replies mentioned in footnotes 135 to 137 in order to support its conclusions in recital 148 of that decision.

204    However, the mere fact that the said replies did not constitute all of the replies received by the Commission does not, in itself, mean that the conclusions drawn in recital 148 of the contested decision are flawed.

205    Google was able, after all, to read all of the replies to the requests for information received by the Commission when accessing the file, including those on which the Commission had based its conclusions in recital 148 of the contested decision. On that occasion, Google was in a position to examine the replies provided by the undertakings and had the opportunity to note that the Commission had at its disposal a larger number of replies from operators than those mentioned in the footnotes presented in recital 148 of the contested decision. This is moreover demonstrated by the fact that it is disputing, in the present plea, the interpretation of certain replies that are not mentioned in that decision. Other than the replies quoted and disputed by Google in the present plea (in paragraphs 168 to 199 above), however, it does not derive any concrete argument from the other replies, which tends to confirm that those replies cannot undermine the Commission’s conclusions in recital 148 of the contested decision, either.

206    For the same reasons, it is appropriate to reject the argument that the number of replies mentioned in the contested decision was, according to Google, limited in so far as it was aware of the fact that the number of undertakings that had replied to the question on prices was higher.

207    Consequently, the Commission could correctly found its conclusions in recital 148 of the contested decision on the replies to the question on prices, even though it failed to cite all of the undertakings’ replies on which it relied.

208    It follows from the foregoing that Google demonstrates neither that the price analysis conducted by the Commission in the contested decision was irrelevant for the purposes of defining the market nor that the conclusions that it drew from that analysis were flawed or misleading.

(e)    Merits of the Commission’s analysis concerning the differences in characteristics between the two types of ad at issue

209    Google claims that the Commission incorrectly analysed the characteristics of the two types of ad at issue and that all of the differences in characteristics between those ads were insufficient to conclude that they lacked substitutability, either from the perspective of advertisers or from that of publishers.

210    It is appropriate to examine, as a first step, Google’s arguments disputing the existence of each of the differences in characteristics put forward in the contested decision, before determining, as a second step, their relevance to the definition of the market.

(1)    Triggering and positioning of the two types of ad at issue

211    In recital 136 of the contested decision, the Commission noted, in relation to the display of the two types of ad at issue, that online search ads appeared only following a keyword query and that they could appear either directly above, directly below or indeed next to the result of such a query. It found, by contrast, that online non-search ads could appear on any website and could be either contextual (that is to say, related to the content of the webpage) or non-contextual (namely, display ads).

212    Google disputes the importance of those differences, arguing that the two types of ad at issue could be displayed on the same pages as search results in interchangeable positions, signifying that different ad spaces are not involved.

213    The Commission disputes Google’s line of argument.

214    In the first place, it is appropriate to note that the findings of the Commission in recital 136 of the contested decision rely on the replies of eight publishers to requests for information and on an extract of the deposition made by a Google representative, [confidential], before the Federal Trade Commission (United States; ‘the FTC’) in May 2012. Moreover, it is apparent also from recital 147 of the contested decision that the World Federation of Advertisers also mentioned that that difference in the positioning of the two types of ad at issue and the fact that online search ads were generated by a user query constituted important points of distinction between the two types of ad at issue. Google, however, criticises neither the accuracy, nor the reliability, nor the consistency of those elements of assessment.

215    In the second place, even though online non-search ads can indeed appear on the same webpages as online search ads, it is not disputed by Google that a user can easily distinguish ads which are related to his or her query from those which are not according to their positioning and content. On the one hand, online search ads are generally found directly below the search bar (and sometimes in a position adjacent to that bar), in list form if the ads are textual and with the express indication that they are ads. On the other hand, their content is directly linked to the keyword query performed by the user. However, online non-search ads may be found in different places on the webpage and the products or services which are advertised therein are not affected by the user’s query.

216    Accordingly, Google does not demonstrate that the Commission’s finding, made in recital 136 of the contested decision and relating to the differences between the two types of ad at issue in terms of triggering and positioning, is flawed.

(2)    Formats of the two types of ad at issue

217    In recital 137 of the contested decision, the Commission noted, in relation to the formats of the two types of ad at issue, that online search advertising was ‘typically exclusively’ text based, whereas online non-search advertising could appear in a variety of textual, graphical and video formats.

218    Google criticises that finding, asserting that, in actual fact, online search ads can also appear in rich formats or include images and that online non-search ads often contain a textual element.

219    The Commission disputes Google’s line of argument.

220    In that regard, it must be pointed out that Google does not deny that the two types of ad at issue tend to appear in different formats. Nor does it dispute that the 13 advertisers and media agencies, as well as the [confidential], to which the Commission refers in recital 137 of the contested decision (see footnote 110), actually highlighted that difference in their replies to the requests for information of the Commission, or that the World Federation of Advertisers had also noted the said difference in a submission of 18 February 2011, as is set out in recital 147 of the contested decision.

221    Furthermore, the screenshot presented by Google in paragraph 54 (Figure 3) of the application in arguing that online search ads often have graphical elements is irrelevant in that regard since it shows another type of online ad, namely the paid specialised search results described in paragraph 33 above. The only issue which Google asks the Court to examine, in the first part of the first plea, is whether the Commission could correctly conclude that online non-search ads were not substitutable for online search ads. It is the only issue raised, both in the heading of the first part of the first plea of the application, which refers exclusively to the analysis of the competition between the two types of ad at issue, and in its content, which in fact pertains only to the Commission’s assessment in recitals 135 to 169 of the contested decision dealing with the same issue. Thus, contrary to what Google argued at the hearing, it did not make any argument in that part of the first plea challenging the Commission’s assessment in recitals 170 to 183 of the contested decision distinguishing online search ads from paid specialised search results.

222    It follows from the foregoing that Google does not present any evidence capable of undermining the finding in recital 137 of the contested decision according to which online search ads and online non-search ads were generally shown in different formats.

(3)    Design costs of the two types of ad at issue

223    In recital 137 of the contested decision, the Commission added that it followed from the abovementioned difference in formats between the two types of ad at issue that advertisers incurred little or no costs when designing online search ads compared to the costs of designing online non-search ads, particularly those containing graphic elements and rich features.

224    Google disputes the veracity of that finding. It asserts that the costs of designing online non-search ads are not necessarily high, recalling that they can also appear in text form or in other, equally very simple formats.

225    The Commission disputes Google’s line of argument.

226    As has been found in paragraphs 220 to 222 above, Google has not put forward evidence capable of calling into question the finding, in the contested decision, that online search ads were typically exclusively text based, whereas online non-search ads often featured a more complex presentation, including a graphic element or a video, while sometimes taking more simple forms. In those conditions, the Commission could consider that the costs of designing online search ads were, as a general rule, lower than those of online non-search ads.

227    Nor has Google criticised the accuracy, reliability or consistency of the replies to the requests for information of four undertakings, mentioned in footnote 112 to the contested decision, on which the Commission relied in support of its finding in recital 137 of the contested decision.

228    Accordingly, Google does not demonstrate that the Commission erroneously found, in recital 137 of the contested decision, that the costs of designing online search ads were generally low compared to the costs of online non-search ads.

(4)    Targeting abilities of the two types of ad at issue

229    In recitals 138 to 141 of the contested decision, the Commission noted that online search ads, being served in response to the user’s keyword query, had an intrinsically higher capability than online non-search ads to answer to an immediate interest of that user. Even though the Commission acknowledged in the contested decision the existence of targeting abilities of certain online non-search ads, such as contextual ads (which adapt to the webpage viewed by the user), behaviourally targeted ads (which adapt to the user’s web history) and ads placed on social networks (which adapt to the user’s social network profile), it took the view that those forms of targeting did not reach the same degree of relevance for the user as online search ads.

230    Google disputes that analysis, arguing that there is no meaningful difference in targeting ability between the two types of ad at issue. It highlights the existence of other products on the market enabling the user to be targeted, such as those marketed by social networks, and retargeting technology enabling the targeting of users who have already visited a website. That evidence was ignored by the Commission, despite the fact that certain advertisers had indicated in their replies to the requests for information that online non-search ads offered highly sophisticated targeting.

231    The Commission disputes Google’s line of argument.

232    In the first place, with regard to Google’s argument that the Commission failed to take account of certain factual elements relevant in its analysis of the targeting abilities of the two types of ad at issue, it should be recalled that the Commission based its conclusion on the superior targeting ability of online search ads, in recitals 138 to 141 of the contested decision, on a range of different elements. Thus, it referred to the replies to the requests for information that it received from 14 undertakings, including advertisers, media agencies and publishers, but also from the [confidential] (footnotes 113 to 115, 119 and 120), and to the information provided by a Google representative before the FTC (see recital 139), to the ‘AdWords Help’ section of Google’s website in 2012 (see recital 140) and to a 2010 report of the Autorité de la concurrence (French competition authority) on online advertising (see recital 141). Moreover, the Commission also noted, in recital 147 of the contested decision, that the World Federation of Advertisers had highlighted the difference in targeting ability between the two types of ad at issue.

233    Google does not make any substantiated argument, however, challenging the accuracy, reliability and consistency of the abovementioned information. Nor does it assert that those elements are irrelevant to the analysis of the targeting abilities of the two types of ad at issue.

234    In the second place, Google does not demonstrate that the Commission failed to examine other relevant items of evidence that might have altered the conclusion of its analysis.

235    First, so far as concerns the targeting abilities offered by the social network operators cited by Google, such as Facebook, LinkedIn and Snap, it must be pointed out that it is not disputed by Google that, in accordance with the Commission’s arguments, the new services marketed by those operators were placed on the market either towards the end of the period concerned by the abuse found in the contested decision, as with Facebook Audience Network in 2014, or after that period, as with LinkedIn Audience Network and Snap Audience Network. Therefore, the first service mentioned above was of limited relevance to the definition of the relevant market, whereas the two others were irrelevant in that regard. Moreover, Google does not dispute the accuracy of the Commission’s finding, in recital 163 of the contested decision, according to which the targeting abilities offered by Facebook Audience Network did not enable the display of ads which answered to the immediate interest of the user as often as online search ads.

236    In addition, Google does not demonstrate that the products mentioned in paragraph 235 above and the other products it refers to in paragraph 10 of Annex C.3 to the reply, namely Criteo, ValueClick and Millennial Media, offered advertisers the possibility of presenting ads which react immediately to an interest expressed by the user, as online search ads do. The reference, in Table 7 of Annex C.3 to the reply, to an 2017 information bulletin of the Federal Cartel Office concerning an investigation into Facebook, in which it is stated that that company was ‘able to improve its targeted advertising activities’, without further detail, cannot prove that that company had during the period of the alleged infringement a product or service liable to target the immediate interest of an internet user in the same way that online search ads would.

237    Second, so far as concerns the retargeting possibilities of certain online non-search ads, which supposedly allow advertisers to target users on the basis of their previous internet usage, it must be noted that the Commission did not ignore that element, contrary to what Google claims. Indeed, in point 2 of recital 141 of the contested decision, the Commission referred to the possibilities for advertisers to target a user who had visited certain websites in the past. However, it considered, while mentioning the replies of six advertisers to the request for information of 11 January 2016 in footnote 120, the interpretation of which is not disputed by Google, that that type of ad, described as ‘behavioural advertising’, was still less likely to correspond to a user’s interest at the moment of exposure and was thus less likely to lead to a purchase or another action on his or her part beneficial to the advertiser.

238    Third, with regard to Commission Decision C(2008) 927 final of 11 March 2008 declaring a concentration to be compatible with the common market and functioning of the EEA Agreement (Case No COMP/M.4731 – Google/DoubleClick) (‘the Google/DoubleClick decision’), on the one hand, it should be recalled that the Commission is not bound by the assessments of the relevant markets carried out in its earlier decisions (see, to that effect, judgments of 14 December 2005, General Electric v Commission, T‑210/01, EU:T:2005:456, paragraphs 118 to 120, and of 11 January 2017, Topps Europe v Commission, T‑699/14, not published, EU:T:2017:2, paragraph 93). The Commission is required to carry out an individual appraisal of the circumstances of each case, without being bound by previous decisions concerning other undertakings, other product and service markets or other geographic markets at different times. Thus, an applicant is not entitled to call the Commission’s findings into question on the ground that they differ from those made previously in a different case, even where the markets at issue in the two cases are similar, or even identical (see judgment of 25 March 2015, Slovenská pošta v Commission, T‑556/08, not published, EU:T:2015:189, paragraph 197 and the case-law cited). On the other hand and in any event, while the Commission did indicate, in recital 52 of the Google/DoubleClick decision, that, according to the replies of advertisers during its market investigation, the targeting abilities of online non-search ads were improving and, in recital 12 of that decision, that they were converging, through ‘behavioural targeting’, with those offered by online search ads, it did not conclude in that decision that it followed that the two types of ad at issue had equivalent targeting abilities.

239    Fourth, so far as concerns the replies of four advertisers to the request for information of 11 January 2016, extracts of which are presented in Table 7 of Annex C.3 to the reply, which illustrate, according to Google, the targeting abilities of online non-search ads, it is indeed true that the replies of three of them, namely [confidential], [confidential] and [confidential], highlight the use of the targeting abilities of online non-search ads. However, those replies do not indicate that the targeting abilities of online non-search ads would allow the internet user’s immediate interest to be answered to and thus encourage him or her to make a purchase, in the same way that online search ads would. Those replies are thus not such as to counterbalance the elements contained in the contested decision, summarised in paragraph 232 above, relating to the differences in targeting abilities between the two types of ad at issue.

240    Accordingly, without it being necessary to rule on the admissibility, disputed by the Commission, of the line of argument of Google contained in Annex C.3 to the reply, it must be concluded that Google does not put forward any elements capable of calling into question the finding of the Commission, in recitals 138 to 141 of the contested decision, that online search ads were more likely to answer to the user’s immediate interest.

(5)    Purposes of the two types of ad at issue

241    In recitals 142 to 144 of the contested decision, the Commission considered, in relation to the purposes of the two types of ad at issue, that online search advertising was more suitable for triggering a purchase, whereas online non-search advertising was more efficient at creating brand awareness.

242    Google claims that no meaningful difference between the purposes of the two types of ad at issue was demonstrated by the Commission and that, in essence, those ads ultimately share the same objective.

243    The Commission disputes Google’s line of argument.

244    In the first place, it must be stated that the Commission founded its assessment in recitals 142 to 144 of the contested decision on a number of elements. Thus, it referred (i) to the replies of publishers and advertisers to its requests for information, relying, in footnotes 122 to 124, inter alia on six replies that it received, (ii) to a March 2010 study carried out by the consultancy firm Econsultancy and the organisation SEMPO and based on a survey of advertisers and media agencies, and (iii) to a September 2008 email of [confidential] at Google.

245    Google, however, makes no argument disputing the accuracy, reliability or consistency of those elements, apart from the allegedly selective nature of the quotation of [confidential]’s email. However, that latter argument is moreover unconvincing, since Google explains only that that person’s email was reacting to an article published in The Wall Street Journal, which had found that the two types of ad at issue were viable means of reaching customers. It thus explains neither how the Commission took the email out of context by quoting it, nor why the statements quoted are unreliable.

246    In the second place, in respect of Google’s argument that the paper of the economist firm RBB on competition from Facebook in online advertising, drafted for Google and dating from November 2016, presented in Annex A.3 to the application, shows that the two types of ad at issue ultimately share the same objective of converting demand into a transaction, it is indeed noted in Section 3.3 of that paper that Facebook allowed advertisers to choose criteria in order to encourage a favourable action on the part of the internet user viewing the ads (for example clicks on the ad to the advertiser’s website). However, that paper does not demonstrate that ads on the Facebook social network have the principal objective of leading the user directly to a purchase by answering to his or her momentary interest in a certain product or service, unlike online search ads.

247    Accordingly, without it being necessary to rule on the admissibility, disputed by the Commission, of the line of argument of Google contained in the reply, it must be concluded that Google’s arguments do not support the finding that the Commission erroneously considered, in recitals 142 to 144 of the contested decision, that the two types of ad at issue had different purposes.

(6)    CTRs and conversion rates of the two types of ad at issue

248    In recital 145 of the contested decision, the Commission noted that online search ads had better CTRs and conversion rates than online non-search ads. In other words, according to the Commission, the odds that the user would click on an ad and proceed to a purchase or to another action bringing value for an advertiser were higher when viewing an online search ad than when viewing an online non-search ad.

249    Google argues that the Commission should have looked at whether the better conversion rate of online search ads found by the Commission was accompanied by a price difference. In that case, online search ads could still be constrained competitively by online non-search ads, meaning that the two types of ad at issue would be part of the same market.

250    The Commission disputes Google’s line of argument.

251    In the first place, it must be observed that Google does not dispute that online search ads have better CTRs and conversion rates than online non-search ads. Google does not therefore call into question the Commission’s finding, made in recital 145 of the contested decision, on the difference between the two types of ad at issue in that respect.

252    In the second place, with regard to Google’s line of argument according to which it was for the Commission to examine whether that difference was directly compensated for by a price difference such that the two types of ad at issue could still be in the same market, it is true that the Commission did not specifically examine that matter in the contested decision, despite the fact – noted by Google – that it had mentioned, in recital 149 of the said decision, the Statista report which expressly remarked on the substantial price difference between the two types of ad at issue. However, the Commission did conduct, in the contested decision, an analysis of the prices of the two types of ad at issue. Indeed, as is described in paragraphs 145 to 208 above, the Commission evaluated whether, from the perspective of publishers, advertisers and media agencies, the two types of ad at issue could be substitutable in the event of a permanent 5-10% increase in the price paid by advertisers for online search ads or an equivalent reduction of the revenues received by publishers for those ads. As was stated in recital 148 of the contested decision, however, a majority of publishers, half of media agencies and all publishers indicated that it was unlikely that such a change would cause them to switch to online non-search ads. As the Commission maintained at the hearing, that analysis also demonstrated that the price for advertisers and media agencies or the revenues from the ads at issue for publishers was not the decisive factor in the choice of operators in that regard, as operators look at the overall value that ads can bring them, which is dictated by a series of factors and not only by price or ad revenue.

253    Accordingly, Google does not demonstrate that the Commission erroneously found, in recital 145 of the contested decision, that there was a difference in the CTRs and conversion rates between the two types of ad at issue.

(7)    Possibilities of measuring the performance of the two types of ad at issue

254    In recital 146 of the contested decision, the Commission noted that online search ads enabled advertisers to estimate return on investment more easily. The Commission explained, in that regard, that it was possible to follow the number of clicks on such ads as well as the purchases made by users following those clicks, which was not possible for online non-search ads, where typically there was no direct connection between the viewing of the ad and the purchase of the product concerned.

255    Google disputes that assessment, maintaining that there are many ways that advertisers can track the performance of online non-search ads in terms of conversion (conversion tracking) and return on investment, including tools that it has introduced to the market. Moreover, the Commission was wrong to cite a study by the consultancy firm Econsultancy, in association with the undertaking ExactTarget.

256    The Commission disputes Google’s line of argument.

257    In the first place, it should be recalled that the Commission based its finding, made in recital 146 of the contested decision and concerning the superior ability of online search ads to measure return on investment, on the replies to requests for information of five advertisers and three media agencies, and of the [confidential], which are mentioned in footnote 132 to that decision. Google, however, makes no argument about those replies and thus disputes neither the accuracy, reliability nor the consistency of those elements.

258    In the second place, concerning Google’s criticism of the reference made by the Commission to the Econsultancy study in recital 146 of the contested decision, it should be pointed out that it is in fact apparent from that study, dating from February 2010 and based on a survey of 1 123 advertisers and media agencies, that the replies of those operators demonstrated that online search advertising was the best ‘channel’ for measuring the return on investment of online search ads. Indeed, it is indicated in Section 4.3.3 (pp. 42 and 43) of that study, presented in Annex B.1 to the defence, that 54% of advertisers and 35% of media agencies that had responded to it indicated that that type of ad was ‘good’ at measuring return on investment, whereas only 37% of advertisers and 23% of media agencies indicated that display advertising was ‘good’ in that regard. Even though, by the question asked in the survey, operators were not asked to compare the two types of ad at issue in that respect, it follows that the replies nevertheless supported the finding of the Commission according to which online search ads enabled advertisers to measure return on investment more easily than online non-search ads. That latter finding is in no way called into question by Google’s argument that 46% of advertisers and 65% of media agencies thought that the ‘digital marketing channel’ in which they were best able to measure return on investment ‘was something other than’ that of online search advertising, since it was not demonstrated that those other channels included online non-search ads. Accordingly, Google’s argument is irrelevant to the matter of the substitutability of the two types of ad at issue.

259    In the third place, so far as concerns Google’s argument that tools were available during the period of infringement for measuring the return on investment of online non-search ads, it is appropriate to note that the information provided by Google shows that those tools granted possibilities for measuring return on investment, but it does not support the conclusion that they provided possibilities in that regard equivalent to those of online search ads.

260    First, with regard to Google Ads, it is apparent from the extracts from its blog, Inside AdWords, that Google’s tools enabled advertisers, in addition to calculating the number of clicks on online non-search ads, to measure the increase in visits to a website and in searches following a display ad campaign. However, it is not indicated that those functionalities enabled advertisers to link their advertising spending to general sales, as is the case for online search ads, for which the Commission indicated in the contested decision – without being contradicted by Google – that it was possible to compare the expenditure linked to a keyword with the purchases resulting from clicks.

261    Second, with regard to the social networks LinkedIn, Twitter and Pinterest and the analytical tools Adobe and Salesforce, which also offer tracking capabilities for online non-search ads, it must be pointed out that, in accordance with what the Commission observes in its written submissions, Google does not state whether those tools were available during the period of infringement or explain how their tracking abilities rivalled those offered by online search ads for measuring return on investment.

262    Third, as far as the Facebook social network is concerned, the RBB paper, presented in Annex A.3 to the application (see paragraph 246 above), describes in Section 3.5 thereof the tools allowing users to be tracked until various types of conversion, including purchase, linked to online non-search ads on that social network. However, irrespective of whether those functionalities were available for advertisers during the period of infringement (see paragraph 235 above concerning Facebook Audience Network), Google does not prove that that Facebook tool was as efficient as online search ads at measuring return on investment, as it does not refer to any of the replies of the advertisers and media agencies to the Commission’s requests for information in that regard.

263    Accordingly, without it being necessary to rule on the admissibility, disputed by the Commission, of the line of argument of Google contained in Annex C.3 to the reply, it must be concluded that Google’s arguments do not demonstrate that the Commission erroneously concluded, in recital 146 of the contested decision, that online search ads enabled advertisers to estimate return on investment more easily than online non-search ads.

(8)    Relevance of the differences in characteristics and uses to the definition of the market

264    It follows from paragraphs 211 to 263 above that Google has not managed to call into question the merits of the Commission’s analysis in the contested decision highlighting the differences between the two types of ad at issue in terms of their triggering and positioning, their formats, their design costs, their targeting abilities, their purposes, their CTRs and conversion rates as well as the possibilities of measuring their return on investment. Those findings were based on various items of evidence the accuracy, reliability and consistency of which Google has not succeeded in calling into question.

265    In that regard, it must be held that, in the overall market definition assessment, carried out by the Commission, those differences constituted relevant indicators that the two types of ad at issue were not substitutable.

266    Indeed, according to the case-law and the market definition notice cited in paragraphs 125 and 126 above, the characteristics and uses of products are relevant to the definition of the market, since a relevant product market comprises all those products which are regarded as interchangeable or substitutable by the consumer, by reason of the products’ characteristics, their prices and their intended use.

267    Moreover, contrary to Google’s arguments, the Commission revealed that the two types of ad at issue did have different purposes, as is apparent from paragraphs 241 to 247 above.

268    However, Google submits that those differences are still insufficient to conclude that the two types of ad at issue were not substitutable in the absence of a substitutability analysis based on the prices of the ads at issue.

269    In that regard, it must be recalled, as has been set out in paragraphs 155 to 161 above, that a price-based substitutability analysis, such as a SSNIP test, was not obligatory as part of the market definition that the Commission carried out. It is required neither by the case-law nor by the Commission in the market definition notice.

270    Moreover and in any event, the Commission analysed in the contested decision whether, from the perspective of publishers, advertisers and media agencies, the two types of ad at issue could be substitutable in the event of a permanent 5-10% increase in the price paid by advertisers for online search ads or an equivalent reduction in the revenues received by publishers for those ads.

271    That analysis allowed the Commission to draw useful learnings for the definition of the relevant market, which are summarised in recital 148 of the contested decision and confirmed in paragraphs 145 to 208 above.

272    Furthermore, as has already been noted in paragraphs 141 to 144 above, it should be recalled that the Commission also relied on other factors in its analysis of the substitutability of the two types of ad at issue, such as the investments necessary for the provision of online search advertising services (recitals 150 to 154 of the contested decision) and the market conduct of certain publishers (recitals 162, 164 and 165 of that decision). It follows that Google cannot criticise the Commission for having founded its market definition analysis solely on the alleged characteristics of the two types of ad at issue.

273    It follows from the foregoing that the differences in characteristics and uses between online search ads and online non-search ads, which were relevant for the definition of the relevant market, were part of a range of evidence that the Commission took into account in its overall market definition assessment.

(f)    Taking into account of examples of the actual conduct of publishers which have allegedly replaced or would replace online search ads with online non-search ads

274    In recitals 162 and 164 of the contested decision, the Commission concluded that the examples of publishers mentioned by Google, such as [confidential], [confidential] and [confidential], which had allegedly replaced online search ads with online non-search ads did not demonstrate that those two types of ad were substitutable.

275    Google maintains that the Commission concluded in the contested decision, wrongly, that the examples of the actual conduct of publishers, such as [confidential], [confidential], [confidential], [confidential] and [confidential], which used or use the two types of ad at issue interchangeably, did not demonstrate that those two types of ad were substitutable. That substitutability is even confirmed by the fact that online search advertising intermediation revenues declined dramatically between 2012 and 2016, whereas advertisers’ total online advertising expenditure increased substantially.

276    Surfboard is of the view that the Commission failed to consider evidence submitted by Google showing substitution by publishers between online search ads and other online ad formats.

277    The Commission disputes Google and Surfboard’s line of argument.

278    In the first place, so far as concerns the examples of [confidential] and of [confidential], it is explained in recital 162 of the contested decision that, in January 2014 and in January 2015 respectively, those operators decreased their use of AFS for their websites on mobile devices and, consequently, online search ads on their websites, following a reduction by Google of the revenue share allocated to those publishers for the display of the ads at issue. It is also apparent from the contested decision that those publishers chose, at a later stage, to increase once again their use of online search ads when they obtained from Google an increase in the percentage of revenues in their favour.

279    As the Commission notes in its written submissions, however, those examples are not such as to demonstrate that the two types of ad at issue were substitutable. First, it is not established that [confidential] and [confidential] used online non-search ads instead of online search ads, following the reduction of the percentage of revenues that were allocated to them. Second, that decrease in the use of AFS occurred after a very significant reduction of the percentage of revenues received by [confidential] and [confidential]. The part of those revenues paid to publishers decreased to [confidential]%, whereas, according to what the Commission explains in its written submissions without being contradicted by Google, it represented between [confidential] and [confidential]% for [confidential] and [confidential]% for [confidential] before that change. Those figures highlight a substantial reduction of publishers’ revenues. Such a reduction is far more significant than that envisaged, as regards a hypothetical price increase, in the context of the SSNIP test, as is described in paragraphs 17 and 18 of the market definition notice.

280    In the second place, so far as concerns the examples of [confidential] and [confidential] presented in Annex A.16 to the application, which are not mentioned in the contested decision, the screenshots presented by Google show that their websites can display, following a query and according to the keywords used, either online search ads or online non-search ads. The publishers concerned therefore chose, according to Google, between the two types of ad at issue depending on the keyword query that the user performed.

281    However, the mere fact that [confidential] and [confidential] displayed the two types of ad at issue on their websites does not in itself mean that those ads are substitutable. Indeed, as the Commission maintains in its written submissions, the choice of those two publishers can be explained simply by the various possibilities for monetising searches through advertising. While certain keywords could encourage advertisers to place online search ads and thereby generate revenues for those two publishers, it could be that other keywords create less or no interest on the part of advertisers, such that publishers generated better revenues by reserving ad space to online non-search ads.

282    In the third place, regarding the example of [confidential], examined in recital 164 of the contested decision, the screenshot presented by Google in Annex A.16 to the application shows that that publisher uses online non-search graphical ads. Google has not demonstrated that that publisher used, in the past, online search ads for the same ad space and had decided to replace them with online non-search ads. Even assuming that to be the case, Google has not produced any element enabling that operator’s choice to be understood, in accordance with what the Commission noted in recital 164 of the contested decision. Accordingly, the screenshot cannot be regarded as demonstrating that the two types of ad at issue are substitutable.

283    In the fourth place, so far as concerns the screenshots of the website ‘[confidential]’ and that of [confidential], presented respectively in Annex A.16 to the application and in Annex C.3 to the reply, it should be noted that those examples show the display of online search ads on the same webpage as that of specialised search results. Those examples are thus irrelevant to the question of whether online non-search ads are substitutable for online search ads. That latter question, however, is the only issue which Google asks the Court to examine in the first part of the first plea, as has been noted in paragraph 221 above.

284    In the fifth place, in terms of the decline in online search advertising intermediation revenues between 2012 and 2016, suffice it to state that Google does not demonstrate the existence of a corresponding increase in the revenues from online non-search ads during that period, such that there may have been a shift to the latter type of ad. In addition, even if such a shift were established, Google does not explain, either, how that fact would be synonymous with substitutability of the two types of ad at issue.

285    Accordingly, without it being necessary to rule on the admissibility, disputed by the Commission, of the line of argument of Google contained in Annex C.3 to the reply, it must be concluded that Google does not demonstrate that the Commission erroneously assessed, in the contested decision, the examples presented by Google of certain publishers which may have used or may use the two types of ad at issue interchangeably. Nor does it establish that the Commission failed to take into account elements relevant to that subject in its substitutability analysis.

(g)    Interpretation of the statements of certain Google representatives

286    In recitals 139, 144 and 156 of the contested decision, the Commission quoted statements made by certain Google representatives citing differences between the two types of ad at issue. In recital 139, in its analysis of the ads’ targeting abilities, it referred to a deposition, before the FTC in June 2012, of [confidential], then [confidential] at Google. In recital 144, in its analysis of the purposes of ads, the Commission quoted an internal email of [confidential] of September 2008 (see, to that effect, paragraph 245 above). In recital 156, in response to an argument of Google on the competitive relationship between the two types of ad at issue, the Commission mentioned the deposition made by [confidential], then [confidential] at Google, before the FTC in May 2012.

287    Google alleges that the Commission took those statements out of context and that they did not address the question of the marginal substitution of the two types of ad at issue.

288    The Commission disputes Google’s line of argument.

289    In the first place, regarding the deposition of [confidential] quoted in recital 139 of the contested decision, it should be noted that that individual clearly indicated in his deposition, an extract of which is presented in Annex A.29 to the application, that online search ads were the best online ads for generating purchases with the best return on investment, since they responded directly to an interest expressed by the user. He asserted, moreover, that advertisers therefore preferred online search ads. He however noted that certain persons disagreed with that latter assertion, namely those who wish to increase brand awareness, an objective for which online non-search ads were more effective.

290    It follows that the Commission’s reference to that deposition, in recital 139 of the contested decision, in the part of its analysis addressing the targeting abilities of the two types of ad at issue, is neither selective nor misleading.

291    In the second place, in respect of the mention of an internal email of [confidential], in recital 144 of the contested decision, presented in Annex A.31 to the application, it is appropriate to note that that email was drafted in response to an article published in The Wall Street Journal in which it had been found that a user was more likely to purchase a product following an online search ad if he or she had already seen display ads (online non-search ads) for the same product. The article stressed therefore that the two types of ad at issue were complementary and not substitutable. In the said email, [confidential] expressed, in essence, his agreement with that analysis, indicating, as is stated in recital 144 of the contested decision, that online non-search ads created the interest in a product, but that online search ads encouraged the purchase of the product.

292    The Commission could therefore correctly quote the comments of [confidential] in its analysis of the purposes of the two types of ad at issue where it found that online non-search ads were more efficient than online search ads at increasing brand awareness. The fact, noted by Google, that [confidential]’s email had been sent in response to an article in which it was observed that online non-search ads also constituted a means for advertisers to reach consumers by no means diminishes the probative value of the opinion expressed by that person and relied on by the Commission in recital 144 of the contested decision.

293    In the third place, in respect of the reference to the quotes of [confidential] contained in recital 156 of the contested decision, it must be recalled that the Commission made that reference to argue that Google itself acknowledged the differences between the two types of ad at issue, particularly in terms of their targeting abilities. It is in fact apparent from that deposition, the complete version of which is presented in Annex B.5 to the defence, that [confidential] made a number of observations relating to certain differences between online search ads and the online search ads shown by Google via its AdSense For Content service, like the fact that the former ads answered to the queries made by users and the different display of ads (pp. 70, 77 and 81). She also noted that, for the purposes of applying the exclusivity clauses in force for the use of AFS, those two types of ad were not regarded as ‘substantially similar’ (pp. 155 and 156).

294    Consequently, while it is indeed true that, in line with Google’s arguments, [confidential] also stated in her deposition that the two types of ad shown via AFS and AdSense For Content were used interchangeably on publishers’ sites (see p. 64 of the said deposition), the fact remains that that Google representative noted that there were differences between those two types of ad, particularly as regards targeting abilities. Therefore, the mention made by the Commission in recital 156 of the contested decision to assert that Google itself acknowledged the differences between the two types of ad at issue is not misleading. It is appropriate to observe, moreover, that that acknowledgement is also confirmed by the statements of [confidential], examined above and mentioned in recitals 139 and 144 of the contested decision.

295    Furthermore, the fact that the quotations of the statements of [confidential] and [confidential] do not refer to the matter of the marginal substitutability of the products from the perspective of advertisers and publishers, as Google alleges, is irrelevant in so far as those quotations were only used by the Commission to illustrate other elements of its overall assessment of the definition of the relevant market, namely the targeting abilities of the two types of ad at issue, their purposes and their differences in characteristics and uses in several respects.

296    It follows from the foregoing that Google does not demonstrate that the Commission misread the statements of its representatives quoted in recitals 139, 144 and 156 of the contested decision.

(h)    Past Commission decisions

297    In recitals 158 and 159 of the contested decision, in response to Google’s arguments, the Commission expressly denied having concluded that the two types of ad at issue were substitutable in merger control decisions that it adopted, including the Google/DoubleClick decision.

298    Google notes that the Commission has acknowledged, in past decisions, that the two types of ad at issue were converging and competed with each other. It refers in that regard to the Google/DoubleClick decision and to Commission Decision C(2010) 1077 final of 18 February 2010 declaring a concentration to be compatible with the common market and the EEA Agreement (Case No COMP/M.5727 – Microsoft/Yahoo! Search Business) (‘the Microsoft/Yahoo! decision’), and Commission Decision C(2010) 5272 final of 27 July 2010 declaring a concentration to be compatible with the common market and the EEA Agreement (Case No COMP/M.5676 – SevenOne Media/G+J Electronic Media  Service/Tomorrow Focus Portal/IP Deutschland/JV) (‘the SevenOne Media JV decision’). It observes that convergence between the two types of ad at issue has increased since the adoption of those decisions.

299    The Commission disputes Google’s line of argument.

300    In the first place, it should be recalled that, in accordance with the case-law cited in paragraph 238 above, the Commission is not bound by the assessments of the relevant markets carried out in its earlier decisions.

301    In the second place and in any event, the Commission committed no error in finding, in recitals 158 and 159 of the contested decision, that its past decisions did not contain conclusions contrary to the definition of the market of online search ads.

302    It should be pointed out, in that regard, that the Commission explicitly decided to leave open the definition of the market in the three decisions cited by Google. Thus, while the Commission indicated, in the Google/DoubleClick decision, that the two types of ad at issue could be considered substitutable ‘to a certain extent’ from an advertiser’s point of view (paragraph 53 of that decision), not only did it not reach a definitive conclusion on such substitutability, but it expressly noted that those two types of ad were ‘entirely different’ from a publisher’s perspective (paragraphs 54 to 56 of the said decision). It is true that, in the Microsoft/Yahoo! decision, the Commission noted that it had received a certain number of responses, during its market investigation, referring to a convergence in certain respects between the two types of ad at issue (paragraph 74 of that decision), but it also found that a ‘significant number’ of responses had emphasised that those ads formed separate markets (paragraphs 71 and 72 of the same decision). Last, in the SevenOne Media JV decision, the Commission limited itself to remarking that the notifying parties had expressed the opinion that the two types of ad at issue were growing closer in terms of targeting abilities (paragraph 30 of that decision) and that the market study had suggested that there might be convergence between those two types of ad (paragraph 31 of that decision).

303    It is appropriate to add, in that regard, that the fact that the Commission noted, in certain decisions, that undertakings had referred to a convergence between the two types of ad at issue in no way means that it itself concluded that those ads were substitutable, contrary to what Google implies. In addition, it should be observed that the Commission did examine, in the contested decision, the potential points of convergence between the two types of ad at issue that were raised in those past decisions, such as the matter of their targeting abilities, only to arrive at the conclusion that they were not part of the same market.

304    It follows from the foregoing that Google is not justified in calling into question the merits of the Commission’s analysis of the substitutability of the two types of ad at issue on the basis of the Commission’s past merger control decisions, to which it makes reference.

(i)    Conclusion on the first part of the first plea

305    It follows from the foregoing considerations that Google’s line of argument does not manage to call into question the accuracy, reliability and consistency of the items of evidence on which the Commission relied in its overall assessment of the substitutability of online search ads and online non-search ads, or to demonstrate that that institution failed to take into account evidence relevant to that end. Accordingly, Google does not demonstrate that the Commission erroneously considered that the two types of ad at issue were not substitutable.

306    In the light of all of the foregoing, the first part of the first plea must be rejected.

2.      Second part of the first plea: substitutability of the sale of online ads via an intermediary and the sale of such ads directly by publishers

307    As has been noted in paragraph 27 above, the Commission defined, in recitals 184 to 200 of the contested decision, a second relevant product market, namely that for online search advertising intermediation, described in paragraph 35 above.

308    In its analysis underpinning the definition of the online search advertising intermediation market, in the first place, the Commission concluded, in recitals 186 to 193 of the contested decision, that there was limited substitutability between the sale of online ads via an intermediary and the sale of online ads made directly by publishers (‘the two sales channels at issue’). First, the Commission justified that conclusion by explaining that intermediated sales did not entail significant costs for publishers, unlike when the latter engage in direct sales. Second, according to the Commission, intermediated sales made it easier to bring together a large number of advertisers, which is indispensable for online advertising, which is why all publishers which sell ads directly also sell via an intermediary. Third, the Commission next examined Google’s arguments on the conduct of certain direct partners, including [confidential] and [confidential], and other publishers, including [confidential], as well as on a past merger control decision, and took the view that those arguments did not allow the distinction between the two sales channels at issue to be called into question (recitals 189 to 193 of the contested decision). In the second place, the Commission concluded, in recitals 194 to 200 of the said decision, that there was limited substitutability between intermediation services for online search ads and those for online non-search ads.

309    Google disputes only the first part of that market definition, by claiming that the Commission erroneously concluded, in the contested decision, that the two sales channels at issue were not substitutable.

310    In the first place, Google considers that the alleged lack of substitutability between the two sales channels at issue was not demonstrated from a publisher perspective. First, it asserts that the Commission did not sufficiently substantiate its findings according to which, on the one hand, direct sales entailed higher transaction costs for publishers than intermediated sales and, on the other hand, intermediated sales allowed a large base of advertisers to be brought together more easily. Second, those two alleged differences in characteristics between the two sales channels at issue do not suffice to conclude that there was no substitutability without having conducted a ‘proper’ assessment of substitutability. Third, the Commission was wrong to consider that the evidence showing that publishers used the two sales channels at issue was insufficient to prove that they were not substitutable.

311    In the second place, Google notes that the Commission rejected, wrongly, the evidence on the substitutability of the two sales channels at issue from the perspective of advertisers.

312    In the third place, Google considers that the Commission was wrong to deem irrelevant Commission Decision C(2012) 6063 final of 4 September 2012 declaring a concentration compatible with the internal market and the functioning of the EEA Agreement (Case COMP/M.6314 – Telefónica UK/Vodafone UK/Everything Everywhere/JV) (‘the Telefónica UK decision’).

313    Surfboard submits that the Commission did not take into account the evidence put forward by Google demonstrating the substitutability of the two sales channels at issue.

(a)    Substitutability of the two sales channels at issue from the perspective of publishers

(1)    Transaction costs for publishers

314    In recital 187 of the contested decision, the Commission concluded that the transaction costs for publishers engaging in direct sales of online ads were higher than those of intermediated sales. After all, in order to sell online ads directly, publishers were required to undertake considerable investments in terms of time, funds and personnel, whereas intermediated sales generated negligible – if not non-existent – transaction costs for them.

315    Google disputes the merits of that analysis. First, it asserts that the observations of the undertakings on which the Commission relied, in recital 187 of the contested decision, were neither conclusive nor reliable. Second, the Commission disregarded the probative value of other evidence indicating that the transaction costs of direct sales were lower than those of intermediated sales. Third, Google notes that the fee paid to intermediaries by publishers, in the form of revenue share from publishers, counteracts any difference in price for publishers.

316    The Commission disputes Google’s line of argument.

317    In the first place, regarding Google’s argument criticising the probative value of publishers’ replies, referred to in footnote 193 contained in recital 187 of the contested decision, on account of the fact that the questions addressed to them did not ask them to compare costs ‘in any meaningful way’, it must be borne in mind that the Commission referred, in the said footnote, to the replies of five publishers to questions set out in its requests for information.

318    Google makes no argument criticising the Commission’s interpretation of those five replies, but criticises the content of the questions asked of three of the five publishers, namely [confidential], [confidential] and [confidential], in the request for information of 22 December 2010. According to Google, those questions did not allow for a proper comparison of the costs incurred by direct sales with those generated by intermediated sales, since their sole aim was to get publishers to provide separately the average of their costs, on the one hand, for direct sales and, on the other hand, for intermediated sales, without comparing those costs.

319    With regard to the questions put by the Commission to the three abovementioned publishers, it should be pointed out that, as Google indicates in paragraph 63 of the application, the said publishers were asked to specify the average costs that they had incurred on an annual basis during the previous six years (from 2005 to 2010) for, on the one hand, direct sales of advertising spaces and, on the other hand, intermediated sales of advertising spaces. They were thus questions of a statistical nature concerning the average of the costs incurred by the two sales channels at issue during the same period. The Commission could therefore, in principle, compare the averages of the costs provided by those publishers. Google makes no substantiated argument, however, explaining why the questions asked were not adequate to obtain information capable of supporting the finding in recital 187 of the contested decision according to which direct sales of online ads incurred higher costs for publishers than intermediated sales. Moreover, it does not argue that another question would have resulted in a conclusion different from the one that the Commission drew on transaction costs.

320    It follows that the accuracy, reliability and consistency of the replies mentioned in footnote 193 are not called into question.

321    In the second place, in terms of Google’s argument that the fee paid to intermediaries counteracts any difference in price for publishers, there is no indication that the fee at issue was not part of the costs incurred by them – as it should have been – and therefore was not included in the average costs calculated by publishers in their replies to the Commission’s requests for information, referred to in paragraphs 318 to 320 above.

322    In the third place, in respect of Google’s argument that the Commission disregarded evidence demonstrating that the costs, for publishers, of direct sales were not higher than those of intermediated sales, Google refers, in its written submissions, only to the reply of [confidential] to the request for information of December 2010. That reply, an extract of which is to be found in Annex A.32 to the application, indicates that [confidential]’s transaction costs for direct sales were lower than those for intermediated sales. Even though the Commission explains, in its written submissions, that [confidential] was an atypical publisher, having developed its own in-house online search advertising services ([confidential]), and having continued to use AFS for online search ads, it is nevertheless an example of a publisher which considered direct sales to entail lower costs than intermediated sales.

323    However, that element alone cannot call into question the finding made by the Commission in recital 187 of the contested decision on transaction costs for publishers. That finding was supported by the replies of five publishers to the requests for information, which were not validly challenged by Google, and, in accordance with the Commission’s arguments, is also consistent with the explanations given in recitals 150 to 154 and 195 to 197 of the contested decision – not disputed by Google – on the significant investments that are necessary to provide online search advertising services.

324    It follows from the foregoing that Google does not demonstrate that the Commission erroneously concluded, in recital 187 of the contested decision, that the transaction costs for the sale of advertising spaces were lower for publishers using an intermediary, in contrast to those engaging in direct sales.

(2)    Indicator relating to access to a sufficient advertiser base

325    In recital 188 of the contested decision, the Commission found that online advertising required a large advertiser base and that intermediated sales allowed publishers to access such a base more easily compared to direct sales of online ads.

326    Google submits that the Commission did not demonstrate that publishers themselves could not bring together a sufficient base of advertisers to sell them online ads directly. To that end, it criticises the interpretation which is made of the reply of the [confidential] to the request for information of 16 September 2011 and maintains that publishers can enjoy an advertiser base large enough to sell ads directly, as publishers such as [confidential] and [confidential] demonstrate.

327    The Commission disputes Google’s line of argument.

328    In the first place, it should be pointed out that, of the evidence mentioned in recital 188 of the contested decision, Google criticises only the Commission’s interpretation of the reply of the [confidential]. It therefore makes no argument challenging the accuracy, reliability or consistency of the replies of the nine publishers to the requests for information referred to by the Commission in footnote 195 in support of its conclusion concerning the advertiser base accessible via intermediated sales of ads.

329    In the second place, concerning the information provided by the [confidential], presented in Annex B.8 to the defence, it is appropriate to note that that replier specified, in reply to Question 3, that the majority of publishers that had responded to its survey used an intermediary to sell advertising space and that all of the publishers that had responded, apart from one which displayed online non-search ads, had indicated that it was unfeasible for them to sell advertising space to advertisers directly, owing inter alia to the lack of profitability, low search-page volumes and the cost of necessary investments. That reply thus shows the strong preference of publishers, according to the survey of the [confidential], for selling ads via an intermediary rather than selling them directly.

330    In the third place, regarding Google’s argument based on the examples of [confidential] and of [confidential], which engage in direct sales, it is necessary to observe that those two examples do not support the view that direct sales generally offer publishers the same size of advertiser base as intermediated sales. As the Commission contends in its written submissions, that factor is particularly important for online search ads, since CTRs – and thus publishers’ rate of remuneration – depend on the relevance of ads relating to the user’s query. The greater the number of advertisers, the more likely it is that the ads will respond to the user’s interest as expressed by his or her query. In addition, those two examples of publishers engaging in direct sales cannot counteract the replies of the nine publishers mentioned by the Commission in footnote 195 to the contested decision, which indicated that the advertiser base made accessible through intermediation was greater than that granted by direct sales and that it was an important factor for publishers.

331    It follows from the foregoing that Google does not demonstrate that the Commission erroneously concluded, in recital 188 of the contested decision, that publishers could have a sufficiently large advertiser base more easily by using intermediated sales for online advertising rather than by making direct sales.

(3)    Lack of ‘proper’ analysis of the substitutability of the two sales channels at issue

332    Google alleges that the Commission could not rely on the evidence relating to transaction costs and the advertiser base accessible to conclude that the two sales channels at issue were not substitutable from a publisher’s perspective without having carried out a ‘proper’ substitutability analysis. Google specified during the hearing that it considered that the Commission ought to have conducted a SSNIP test.

333    The Commission disputes Google’s line of argument.

334    It is appropriate to examine first of all the Commission’s argument in its defence according to which it performed a price analysis. The Commission refers, in that regard, to the fact that it put to publishers, in its requests for information of 26 July 2013 and 18 March 2016, the question whether they would replace all or part of the intermediated sales of online search ads with direct sales of such ads in the event of a significant and non-transitory 5-10% increase in the price of intermediation services. Annexes B.6 and B.7 to the defence in fact contain the replies of 12 publishers to that question. The Commission adds, referring to recitals 187 and 188 of the contested decision and to footnotes 193 and 195 thereto, that all of the publishers that had replied to that question stated that they would be unlikely to make such a switch.

335    However, it must be pointed out in that regard that the contested decision contains no reference to that analysis.

336    While the Commission did mention, in footnotes 193 and 195, certain replies of the publishers to the question indicated in paragraph 334 above and asked in the requests for information, those replies are included in the contested decision not to describe the results of a price analysis, but solely to justify the Commission’s conclusions on transaction costs (replies mentioned in footnote 193 contained in recital 187) and on the accessible advertiser base (replies mentioned in footnote 195 contained in recital 188).

337    It is apparent from the case-law, however, that the Commission cannot supplement the statement of reasons for the contested decision during the proceedings (see, to that effect, judgment of 15 December 2021, Oltchim v Commission, T‑565/19, EU:T:2021:904, paragraph 275 and the case-law cited).

338    That being so, the absence of such an analysis in the contested decision cannot invalidate the definition of the relevant market carried out by the Commission, particularly as regards the substitutability of the two sales channels at issue from a publisher’s perspective.

339    First, it should be emphasised that carrying out a SSNIP test is not obligatory for the purposes of defining the market, as has been explained in paragraphs 155 to 161 above.

340    Second, Google makes no argument in its written submissions alleging that such a test would have resulted in a conclusion different from the one that the Commission drew in the contested decision. It is important to emphasise, in that respect, that the Commission relied, in order to define the relevant market, on a range of evidence relating to the transaction costs for publishers and to the advertiser base accessible to publishers, whose merits and relevance to the substitutability analysis have not been called into question, as has been concluded in paragraphs 314 to 331 above. Moreover, its analysis underpinning the market definition for online search advertising intermediation was also based on considerations which have not been challenged by Google, such as the lack of substitutability between online search advertising intermediation and online non-search advertising intermediation. Given those relevant and consistent elements, the mere fact that the Commission did not conduct a SSNIP test for the purposes of defining the relevant market does not demonstrate that its overall assessment in that regard is erroneous.

341    It follows from the foregoing that Google does not demonstrate that the Commission’s assessment of the market definition was erroneous on the sole ground that it did not conduct a SSNIP test.

(4)    Publishers using the two sales channels at issue

342    In recital 191 of the contested decision, the Commission concluded that the screenshots of the website of [confidential] and the replies of [confidential] and of [confidential], concerning the alleged substitution by those publishers between the two sales channels at issue, were not demonstrative of the substitutability of those channels.

343    Google, supported by Surfboard, claims that the Commission ignored, wrongly, evidence showing that publishers displayed online ads both in direct contact with advertisers and via intermediation. Google cites, to that end, the examples of [confidential] and of [confidential] which demonstrate the substitutability of the two sales channels at issue.

344    The Commission disputes Google’s line of argument.

345    In the first place, it should be noted that Google does not dispute the Commission’s analysis, contained in recital 191, point 2, of the contested decision, of the replies of [confidential] and of [confidential], according to which those replies do not show that the two sales channels at issue are substitutable.

346    In the second place, regarding the example of [confidential], Google presents in Annex A.16 to the application, Figures 6 and 7, two screenshots of webpages following searches on [confidential], which were examined in recital 191, point 1, of the contested decision. Google explains that, whereas Figure 6 shows online search ads generated by Google’s intermediation services, Figure 7 shows an ad of [confidential] (house ad) sold directly by the latter.

347    It should be noted in that regard that the fact that [confidential] uses the two sales channels at issue to sell advertising space does not necessarily demonstrate that those channels are substitutable. It is true that it proves that the same publisher can choose to show its own ads on a page of its website and display online search ads sold via an intermediary on another page. However, in accordance with the Commission’s explanation in recital 191, point 1, paragraph 1, of the contested decision, that choice could be made on the basis of the monetisation possibilities of the keywords used in search queries. In that case, the mere fact that a publisher sells ads both directly and via an intermediary does not demonstrate that the two sales channels at issue are interchangeable or substitutable by reason of their characteristics, their prices and their intended use within the meaning of the case-law and the market definition notice cited in paragraphs 125 and 126 above.

348    In the third place, in respect of the example of [confidential], Google presents, in Annex A.22 to the application, a screenshot following a search query on the website [confidential] showing the display of ads sold directly by [confidential] above the search results and online search ads generated by Google’s intermediation services below the said results.

349    It must be observed, however, that that screenshot was not examined in the analysis of the definition of the relevant market carried out in the contested decision and that there is no indication in the file that Google presented it to the Commission during the administrative procedure. Accordingly, Google cannot criticise the Commission for having ignored that example, contrary to its assertion in the application. In addition, in accordance with the arguments of the Commission in its written submissions, that screenshot, showing ads sold by various sales channels on the same webpage, tends to demonstrate the complementarity between the two sales channels at issue rather than their substitutability. That is why the example of [confidential] is consistent with the publisher conduct described in recitals 188, 190, 191 and 192, point 2, of the contested decision. In those recitals, the Commission noted that several publishers, such as members of the [confidential] and certain of Google’s direct partners, including [confidential], [confidential] and [confidential], sold advertising space directly, but had also systematically used intermediated sales in view, in essence, of the advantages offered by that latter sales channel. Such conduct is likely to be representative of complementarity between the two sales channels at issue rather than substitutability between them.

350    In any event, even assuming that the examples of [confidential] and [confidential] constituted evidence of the existence of substitutability between the sale of online advertising via an intermediary and direct sale by publishers, which has not been demonstrated, it must be held that those elements do not suffice to counteract the evidence presented in recitals 187 and 188 of the contested decision indicating that publishers generally do not regard the two sales channels at issue as substitutable owing to the significantly lower transaction costs of intermediated sales and the easier access to a large advertiser base that intermediation offers compared to direct sales.

351    It follows from the foregoing that Google does not demonstrate that the Commission erroneously rejected, in recital 191 of the contested decision, examples of publishers using the two sales channels at issue in its analysis of the substitutability of those sales channels.

(b)    Substitutability of the two sales channels at issue from the perspective of advertisers

352    In recital 192 of the contested decision, the Commission considered, in essence, that the examples, mentioned by Google, of publishers having developed advanced targeting capabilities linked to user behaviour, namely [confidential], [confidential] and [confidential], gave no indication about the substitutability of the two sales channels at issue.

353    Google argues that the Commission wrongly dismissed that evidence, which shows that the two sales channels at issue are substitutable from the perspective of advertisers. After all, publishers are capable of displaying directly sold ads very effectively in view of the fact that certain of them can collect user data (like [confidential]), in particular through user registration and the use of cookies (like [confidential]), and target users who have already visited a website (like [confidential]).

354    The Commission disputes Google’s line of argument.

355    In that regard, it should be pointed out that the examples presented by Google highlight the user-targeting capabilities offered by online non-search advertising on the websites of certain publishers. Even though those elements could demonstrate that those publishers are capable of selling attractive advertising space to advertisers directly, Google does not explain how that fact alone shows that advertisers regard the placement of online ads directly with publishers as substitutable for the placement of such ads by an intermediary – for example Google and its AFS service as regards online search ads.

356    Moreover, Google does not criticise the Commission’s observation, in recital 192, paragraph 2, of the contested decision, according to which [confidential] and [confidential] had continued to use Google’s intermediation services for online search ads, despite the development of their own tools to target better ads promoting their own products. That finding tends to indicate that [confidential] and [confidential] considered that their tools could not replace intermediated sales.

357    It follows from the foregoing that Google does not demonstrate that the Commission erroneously rejected its examples in recital 192 of the contested decision.

(c)    Taking into account of the Telefónica UK decision

358    In recital 193 of the contested decision, the Commission asserted that the Telefónica UK decision did not support Google’s arguments, since, in that decision, the Commission had left open the question whether the constraint exerted by direct sales of mobile advertising on intermediated sales of that advertising justified enlarging the market at issue.

359    Google claims that the Commission was wrong to dismiss its conclusion in the Telefónica UK decision according to which direct sales of mobile advertising constrained the sale of mobile advertising via an intermediary to a significant extent.

360    The Commission disputes Google’s arguments.

361    In that regard, it should be recalled, as has already been noted in paragraphs 238 and 300 above, that the Commission is not bound by the assessments of the relevant markets carried out in its earlier decisions.

362    Moreover and in any event, it should be observed that the Commission left open, in the Telefónica UK decision, the question of segmentation of the market for the sale of mobile ad space between direct sales and intermediated sales, in accordance with the finding in recital 193 of the contested decision. It specifically indicated that its market study did not allow a clear conclusion on that matter to be drawn.

363    Moreover, Google does not explain, in its written submissions, why the Commission’s observations in the Telefónica UK decision on the possible substitutability of the sales channels for mobile ad space would be relevant to the assessment, in the contested decision, of the substitutability of the two sales channels at issue, which concern online ads.

364    It follows from the foregoing that Google does not demonstrate that the Commission committed an error, in recital 193 of the contested decision, in considering that the Telefónica UK decision did not support its line of argument.

(d)    Conclusion on the second part of the first plea

365    It follows from the foregoing considerations that Google’s line of argument does not manage to call into question the accuracy, reliability and consistency of the items of evidence on which the Commission relied in its overall assessment of the substitutability of the sale of online ads via an intermediary and the sale of online ads directly by publishers, or demonstrate that that institution failed to take account of evidence relevant to that end. Accordingly, Google does not demonstrate that the Commission erroneously considered that the two sales channels at issue had a limited degree of substitutability.

366    In the light of all of the foregoing, the second part of Google’s first plea must be rejected.

3.      Conclusion on the first plea

367    It is apparent from the foregoing considerations that Google has not demonstrated that the definition of the relevant markets carried out by the Commission was erroneous.

368    Consequently, Google’s argument that the Commission did not prove that it had a dominant position on the online search advertising intermediation market in the EEA is also unfounded to the extent that that argument is based solely on the allegedly erroneous definition of that market.  Google’s line of argument is therefore incapable of calling into question the conclusion of the Commission, contained in recital 274 of the contested decision, according to which Google held a dominant position from 2006 to 2016 on the online search advertising intermediation market in the EEA.

369    Accordingly, Google’s first plea must be rejected as unfounded.

C.      Second plea: the exclusivity clause in GSAs concluded with all sites direct partners did not constitute an abuse of a dominant position

370    By its second plea, Google criticises the Commission for having found that the exclusivity clause in GSAs concluded with all sites direct partners constituted an abuse of a dominant position within the meaning of Article 102 TFEU. This plea consists of three parts, the first alleging that that clause did not constitute an exclusive supply obligation within the meaning of the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), the second alleging that the Commission was required to analyse the effects of the clause, and the third alleging that the contested decision does not establish that the clause in question was capable of restricting competition.

371    As a preliminary point, it must be recalled that, in the pre-March 2009 template GSA, the exclusivity clause was worded as follows:

‘For each Agreement Customer agrees that during the applicable Services Term Customer shall not implement on the applicable Site or provide access through the applicable Customer Client Application (if any) any services which are the same as or substantially similar to any of the Services being supplied by Google under the Agreement or which are otherwise directly competitive to such Services.’

1.      First and second parts of the second plea: the exclusivity clause in GSAs concluded with all sites direct partners did not constitute an exclusive supply obligation within the meaning of the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36) and the Commission was required to analyse the effects of that clause

372    First, in the contested decision, the Commission recalled the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), according to which ‘an undertaking which is in a dominant position on a market and ties purchasers – even if it does so at their request – by an obligation or promise on their part to obtain all or most of their requirements exclusively from the said undertaking abuses its dominant position within the meaning of Article [102 TFEU]’ (see paragraph 89 of the judgment).

373    Second, the Commission found that the exclusivity clause constituted, in the present case, an exclusive supply obligation because it obliged all sites direct partners to source all or most of their requirements in terms of online search advertising intermediation services from Google. In that regard, it noted, first, that the exclusivity clause in GSAs concluded with those direct partners applied ‘typically’ to all of their websites displaying online search ads, next, that the said direct partners could not derogate from that clause before the end of their GSAs and, last, that the GSAs concluded with [confidential] and [confidential] required them to make all of their websites displaying such ads subject to the said clause.

374    In those conditions, the Commission considered, primarily, that the exclusivity clause in GSAs concluded with all sites direct partners was contrary to Article 102 TFEU, without it having been required to verify whether that clause was capable of restricting competition in the light of all the circumstances of the case.

375    Google criticises the Commission for having considered that it had not been required to verify that the exclusivity clause in GSAs concluded with all sites direct partners had been capable of restricting competition. It argues, in that regard, that the Commission did not establish that those direct partners were under an exclusive supply obligation which was contrary to Article 102 TFEU pursuant to the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36). In addition, it maintains that the principles laid down by the Court of Justice in the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632), required the Commission to examine the effects of the said clause in order to determine whether it constituted an infringement of that provision.

376    Surfboard criticises the Commission for having found that it was under an exclusive supply obligation.

377    The Commission contends that, once the GSA was concluded, the exclusivity clause that it contained applied for the duration of that GSA. Consequently, a direct partner that had included all of its websites in its GSA was required, after the entry into force of that GSA, to source all of its requirements in terms of online search advertising intermediation services, which constituted an abuse of a dominant position pursuant to the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), exclusively from Google.

378    In addition, the Commission contends that the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632), clarifies the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), only so far as concerns the exclusive supply obligations imposed by a dominant undertaking in consideration of the grant of a rebate or payment, which is not the case here.

379    In the first place, the Court did hold, in the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), that clauses by which contracting parties undertook to purchase all or a considerable part of their requirements from an undertaking in a dominant position, even if not accompanied by rebates, constituted, by their very nature, an exploitation of a dominant position and that the same was true of the loyalty rebates granted by such an undertaking (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 46).

380    However, in the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632, paragraph 138), the Court clarified, in the first place, the abovementioned case-law in a situation where an undertaking in a dominant position submitted, during the administrative procedure, with evidence in support of its claims, that its conduct was not capable of restricting competition and, in particular, of producing the alleged exclusionary effects (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 47).

381    In that regard, the Court stated that, in that situation, the Commission is not only required to analyse, first, the extent of the undertaking’s dominant position on the relevant market and, secondly, the share of the market covered by the challenged practice, as well as the conditions and arrangements for granting the rebates in question, their duration and their amount; it is also required to assess the possible existence of a strategy aiming to exclude competitors that are at least as efficient as the dominant undertaking from the market (judgments of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 139, and of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 48).

382    The Court added, in the second place, that the analysis of the capacity to foreclose is also relevant in assessing whether a system of rebates which, in principle, falls within the scope of the prohibition laid down in Article 102 TFEU may be objectively justified. In addition, the exclusionary effect arising from such a system, which is disadvantageous for competition, may be counterbalanced, or outweighed, by advantages in terms of efficiency which also benefit the consumer. That balancing of the favourable and unfavourable effects of the practice in question on competition can be carried out only after an analysis of the intrinsic capacity of that practice to foreclose competitors which are at least as efficient as the dominant undertaking (judgments of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 140, and of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 49).

383    It is true that, in providing that second clarification, the Court referred only to rebate schemes. However, since both rebate practices and exclusivity clauses are capable of being objectively justified or of having the disadvantages which they generate counterbalanced, or even outweighed, by advantages in terms of efficiency which also benefit the consumer, such a clarification must be understood as applying to both of those practices (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 50).

384    Incidentally, in addition to the fact that such an interpretation appears to be consistent with the first clarification provided by the Court in the judgment of 6 September 2017, Intel v Commission (C‑413/14 P, EU:C:2017:632, paragraph 139), it must be held that, although, by reason of their nature, exclusivity clauses give rise to legitimate concerns of competition, their ability to exclude competitors is not automatic, as, moreover, is illustrated by paragraph 36 of the Communication from the Commission entitled ‘Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings’ (OJ 2009 C 45, p. 7) (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 51).

385    Thus, first, where the Commission suspects that an undertaking has infringed Article 102 TFEU by using exclusivity clauses, and where that undertaking disputes, during the procedure, the specific capacity of those clauses to exclude equally efficient competitors from the market, with supporting evidence, it must ensure, at the stage of classifying the infringement, that those clauses were, in the circumstances of the case, actually capable of excluding competitors as efficient as that undertaking from the market (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 52).

386    Second, the Commission is also required to assess, specifically, the ability of those clauses to restrict competition where, during the administrative procedure, the undertaking which is under suspicion, without formally arguing that its conduct was incapable of restricting competition, maintains that there are justifications for its conduct (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 53).

387    In the case at hand, on the one hand, it is settled that Google had disputed, during the administrative procedure and with supporting evidence, the capability of the exclusivity clause in GSAs concluded with all sites direct partners to restrict competition. On the other hand, it is also settled that Google had maintained that that clause was objectively justified.

388    In those conditions, it was for the Commission to demonstrate that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition and it had to take into account, to that end, all the relevant circumstances of the case.

389    Consequently, it should be noted that, contrary to what it asserted in the contested decision, the Commission could not limit itself to finding, in order to establish an infringement of Article 102 TFEU, that the exclusivity clause in GSAs concluded with all sites direct partners required them to source all or most of their requirements in terms of online search advertising intermediation services exclusively from Google. It also had to demonstrate that the said clause was capable of restricting competition, taking into account all the relevant circumstances of the case, which it incidentally did, in the alternative, in the contested decision.

390    Thus, without it being necessary to rule on the question, raised by Google, of whether the exclusivity clause in GSAs concluded with all sites direct partners did in fact constitute an exclusive supply obligation, like the one referred to by the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), it must be concluded that the Commission was wrong to consider, primarily, that it had not been required to verify whether that clause could restrict competition in the light of all the circumstances of the case.

391    Accordingly, the first and second parts of the second plea must be upheld.

2.      Third part of the second plea: the contested decision does not establish that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition

392    In the contested decision, the Commission considered, in the alternative, in recital 362 thereof, that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition, in the light of all the circumstances of the case. It noted in that regard that that clause had (i) deterred those direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers. In addition, it found that the English clause had exacerbated the capability of the exclusivity clause to restrict competition.

393    More specifically, it should be noted that, in finding that the exclusivity clause in GSAs concluded with all sites direct partners had, on the one hand, deterred those direct partners from sourcing from Google’s competing intermediaries and, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, the Commission essentially considered that that clause was capable of producing a foreclosure effect.

394    In addition, it should be noted that the Commission inferred from the foreclosure effect of the exclusivity clause in GSAs concluded with all sites direct partners that that clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

395    First, it follows from recitals 404 to 406 of the contested decision that the foreclosure effect of the exclusivity clause in GSAs concluded with all sites direct partners deterred Google’s competing intermediaries from providing or developing different online search ads, with the result that that clause deterred them from investing in innovation. Next, it follows from recital 408 of that decision that that effect deprived the said intermediaries of revenues and data that they could have used to provide online search ads. Last, it follows from recital 417 of the same decision that the said effect allowed Google to set the prices paid by advertisers at a high level, thereby increasing the prices consumers paid for the goods featured in the online search ads. The Commission added, in recital 418 of the same decision, that the fact that the said clause had possibly deterred innovation had also deprived consumers of a choice of a wider choice of online search ads.

396    Google submits that the Commission did not establish that the exclusivity clause in GSAs concluded with all sites direct partners, first, had produced the foreclosure effect identified in the contested decision, second, had helped it to maintain or strengthen its dominant position on the national markets for online search ads at issue and, third, had deterred innovation or harmed consumers. In addition, it takes issue with the Commission for not having shown that the English clause was capable of restricting competition.

397    Surfboard claims that the GSAs that it had concluded with Google had not prevented it from sourcing from Google’s competing intermediaries and disputes the fact that online contracts did not allow publishers’ requirements for at least some of their websites to be met. It adds that the exclusivity clause in its GSAs was, in any event, objectively justified.

398    It is appropriate from the outset to examine the foreclosure effect, identified in the contested decision, resulting from the exclusivity clause in GSAs concluded with all sites direct partners. It should therefore be ascertained whether that clause was capable, on the one hand, of deterring those direct partners from sourcing from Google’s competing intermediaries and, on the other hand, of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation.

399    In that regard, it should be noted that, in recital 364 of the contested decision, the Commission specified that, for the purposes of its analysis seeking to demonstrate that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition, it had taken into account all the relevant circumstances, including, on the one hand, the extent of Google’s dominant position, both on the national markets for online search ads at issue and on the market for online search advertising intermediation, and, on the other hand, the share of the latter market covered by the said clause and the ‘duration of [that c]lause’. It referred, in that regard, respectively, to Section 7 of that decision, relating to Google’s dominant position, the content of which is summarised in paragraphs 39 to 66 above, and to the whole of Section 8.3.4.2 of the said decision, relating to the impossibility for Google’s competing intermediaries to access a significant part of the said market.

400    It should be noted that the approach followed by the Commission conforms to the case-law according to which, in the case where an undertaking submits, during the administrative procedure, on the basis of supporting evidence, that its conduct was not capable of restricting competition and, in particular, of producing the alleged foreclosure effects, the Commission is required to analyse, inter alia, the extent of the undertaking’s dominant position on the relevant market, the share of the market covered by the challenged practice, as well as the conditions and arrangements of the clause at issue and their duration (see, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 139).

401    In that regard, it must be noted that, in Section 7 of the contested decision, the Commission considered that Google was in a dominant position inter alia on the market for online search advertising intermediation in the EEA from 2006 to 2016, in view of its market shares, the barriers to entry and expansion and the lack of countervailing buyer power on the part of publishers.

402    First, in regard to Google’s market shares on the market for online search advertising intermediation in the EEA, by relying on its gross revenues, the Commission found, on the one hand, on the basis of data provided by Google, that it had held market shares that were always above [confidential]% between 2006 and 2016 and that had reached, in that latter year, [confidential]% and, on the other hand, on the basis of data provided by Google, Microsoft and Yahoo!, that Google had held market shares above [confidential]% in 2006 and which had always been above [confidential]% between 2007 and 2014. By relying on its net revenues, the Commission found, on the one hand, on the basis of data provided by Google, that it had held market shares above [confidential]% in 2006 and above [confidential]% between 2007 and 2016 and, on the other hand, on the basis of data provided by Google and Yahoo!, that Google had held market shares that were always above [confidential]% between 2006 and 2011 and which had reached, during that latter year, over [confidential]%. The Commission inferred from this that Google had faced very limited competition from other intermediaries.

403    Second, the Commission considered that there were numerous barriers to entry and expansion in the market for online search advertising intermediation in the EEA. In that regard, it stated, inter alia, that significant investments were required in order to establish, maintain and refine a ‘search advertising platform’ and that the online search advertising intermediation market was characterised by network effects. It noted that the success of an intermediary depended on the number of advertisers and publishers that it could attract as well as the size of its portfolio of online search ads. Thus, the greater the number of advertisers that used an online search advertising intermediation service, the more ads related to those searches the intermediary could choose from and thus increase the relevance of the ads that it served in response to a user’s query.

404    Third, the Commission considered that the online search advertising intermediation market in the EEA was characterised by a lack of countervailing buyer power on the part of publishers.

405    Google does not challenge the content of Section 7 of the contested decision other than by arguing, under the first plea, that the Commission erroneously defined the relevant markets in Section 6 of that decision.

406    Furthermore, although the ability of exclusivity clauses to exclude competitors is not automatic, as, moreover, is illustrated by paragraph 36 of the Communication from the Commission entitled ‘Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings’, the fact remains that, by reason of their nature, those clauses give rise to legitimate concerns of competition (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 51).

407    In the case at hand, it is settled that, as is apparent from its wording in the pre-March 2009 template GSA, the exclusivity clause prohibited direct partners from displaying competing ads on the websites included in a GSA, for the duration of that agreement. There were no exceptions to that prohibition.

408    Taking into account those factors, it is appropriate to examine, in the first place, whether the exclusivity clause in GSAs concluded with all sites direct partners was capable of deterring them from sourcing from Google’s competing intermediaries and, in the second place, whether the said clause was capable of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation.

(a)    Deterrent effect of the exclusivity clause vis-à-vis all sites direct partners

409    In the contested decision, the Commission found that the exclusivity clause had deterred all sites direct partners from sourcing from other intermediaries in order to display competing ads on their websites or on certain of their pages.

410    Google claims, in essence, that the choice of all sites direct partners typically to include all of their websites in their GSAs resulted from competition on the merits, meaning that they would not have sourced from other intermediaries, even if there was no exclusivity clause in those GSAs. It cites, in that regard, (i) the fact that those direct partners were, in principle, free to choose the websites that they included in their GSAs, (ii) the replies of the said direct partners to the Commission’s requests for information, (iii) the choice and the amount of the investments made by Yahoo!, and (iv) a study that it produced during the administrative procedure.

411    As a preliminary point, it should be recalled that, in accordance with the case-law cited in paragraph 108 above, in order to establish that the exclusivity clause was abusive vis-à-vis all sites direct partners, the Commission did not necessarily have to demonstrate that that clause had actually produced anticompetitive effects. In order to find an infringement of Article 102 TFEU, it is sufficient for the Commission to establish that the said clause had, during the period in which it was implemented, the ability to restrict competition. It follows that the Commission was not required, in the contested decision, to demonstrate that the exclusivity clause had actually deterred each and every all sites direct partner from sourcing from Google’s competing intermediaries.

(1)    Choice of direct partners to include a website in a GSA

412    Google claims that direct partners could choose the websites that were included in their GSAs. Thus, it argues that, in so far as direct partners were not, in principle, obliged to include all of their websites in their GSAs, the exclusivity clause had not been able to deter them from sourcing from another intermediary. In that regard, it recalls that direct partners could use AFS by means of online contracts and that they could, at any time, opt to use a competing online search advertising intermediation service on the websites that were not included in their GSAs.

413    The Commission disputes Google’s line of argument.

414    In that regard, it is settled that, pursuant to the exclusivity clause, a direct partner could not, in principle, display competing ads on websites which were included in a GSA. It follows that once a direct partner had chosen to include one of its websites in a GSA, it necessarily had to source its requirements in terms of online search advertising intermediation services exclusively from Google as far as that website was concerned.

415    In those conditions, it must be held that the Commission was right to consider that the exclusivity clause could deter direct partners from displaying competing ads on the websites included in their GSAs, notwithstanding the fact that they could conclude online contracts and choose websites which were included in those GSAs.

(2)    Replies of direct partners to the Commission’s various requests for information and Surfboard’s letter

416    In the first place, in recital 348 of the contested decision, the Commission listed the various legal entities that together made up the all sites direct partners. It thus identified 34 all sites direct partners.

417    In the second place, on the one hand, in recital 367 of the contested decision, the Commission quoted the replies of seven direct partners to a request for information of 22 December 2010 to find that all sites direct partners would have sourced at least part of their requirements from other intermediaries were it not for the exclusivity clause.

418    The seven direct partners identified by the Commission in recital 367 of the contested decision are [confidential] (one of whose subsidiaries has been, since October 2010, [confidential] which is one of the all sites direct partners mentioned in recital 348 of that decision), the [confidential] group (to which [confidential] belongs), [confidential] (whose group owns [confidential] and [confidential], which together constitute one of the all sites direct partners within the meaning of that same recital 348), the [confidential] group (to which belong [confidential] and [confidential], which each constitute separate direct partners within the meaning of the said recital 348), the [confidential] group (to which belong [confidential] and [confidential], which each constitute separate direct partners within the meaning of recital 348 of the contested decision), the [confidential] group (to which belongs [confidential]) and [confidential].

419    On the other hand, in recital 368 of the contested decision, the Commission quoted the replies of two direct partners to find that the exclusivity clause had prevented all sites direct partners from evaluating the commercial interest in sourcing from competing intermediaries.

420    The two direct partners identified by the Commission in recital 368 of the contested decision are the [confidential] group, already mentioned in recital 367 of that decision, and the [confidential] group.

421    With the [confidential] group being mentioned in both recital 367 and recital 368 of the contested decision, it should be noted that, in all, the Commission mentioned the replies of eight separate direct partners in those recitals. It follows from recital 348 of the contested decision that those 8 direct partners in reality represent 10 of the 34 all sites direct partners identified by the Commission.

422    Google notes that, in the contested decision, the Commission took the view that the exclusivity clause had deterred the entirety of the all sites direct partners identified in the said decision from sourcing from other intermediaries. It observes however that that decision refers only to the statements of only some of those direct partners. Moreover, it argues that, taken as a whole, the statements of the direct partners show, first, that they were not affected by the exclusivity clause and, second, that their choice to use AFS resulted from competition on the merits, that is to say, that they chose AFS because it was better than competing services.

423    Surfboard adds that the exclusivity clause did not prevent it from sourcing part of its requirements from one of Google’s competitors.

424    The Commission contends that it relied on the replies of 8 of the 34 all sites direct partners that it identified, whereas Google refers, in the application, only to the replies provided by only 2 of those direct partners. In addition, it notes that the reply of one of those two direct partners corroborates the fact that all sites direct partners would have sourced at least part of their requirements from other intermediaries were it not for the exclusivity clause. As for the other replies provided by the other direct partners cited by Google in Annex C.1 to the reply, either they are irrelevant, since they are from direct partners that are not all sites direct partners, or they do not corroborate the fact that those direct partners chose AFS, because it was better than competing services. In addition, the Commission notes that 24 of the 35 replies quoted in the said Annex C.1 pertained to Question 5.2.d of the request for information of 22 December 2010. That question, however, concerned the change of intermediaries in general, and not specifically the quality of AFS compared to that of those intermediaries’ services.

425    Last, the Commission contends that Surfboard does not show that it did not wish to source from another intermediary.

(i)    Relevance of the replies to Question 5.2.d of the request for information of 22 December 2010

426    In support of its line of argument recalled in paragraph 422 above, Google cites the replies of direct partners to Question 5.2.d of the request for information of 22 December 2010.

427    In that regard, it should be noted that Question 5.2.d of the request for information of 22 December 2010 was worded as follows: ‘In which circumstances [would you] consider switching for part or all of your advertising space to a different intermediary?’

428    Thus, as the Commission maintains, Question 5.2.d concerns the change of intermediaries in general, and not specifically the quality of AFS compared to that of competing services.

429    However, it cannot be ruled out that, in replying to Question 5.2.d, direct partners were able to mention factors relevant to their decision to use AFS and, as the case may be, the role that the exclusivity clause might have played in that decision. Furthermore, first, it must be pointed out that the Commission itself relied, in paragraph 142 of the defence, on a reply to that same question in order to counter one of Google’s arguments. Second, it must be noted that the Commission referred, in recitals 367 and 368 of the contested decision, to replies from the [confidential] group to Questions 5.2.c and 5.2.e which, like Question 5.2.d, concern, in a general manner, the considerations relating to the choice to source from a given intermediary and not specifically the quality of AFS compared to that of competing services.

430    In those conditions, contrary to what the Commission suggests, it is impossible to conclude that the replies of the direct partners to Question 5.2.d are irrelevant before examining them individually.

(ii) Relevance of the replies of direct partners that are not all sites direct partners

431    In support of its line of argument recalled in paragraph 422 above, Google cites the replies to various requests for information of the Commission from direct partners which had not been classified as all sites direct partners in the contested decision.

432    The Commission disputes Google’s line of argument.

433    In that regard, as has been noted in paragraph 67 above and as is apparent inter alia, in essence, from recitals 341, 362, 366, 380, 403, 407, 416, 422, 627 and 630 of the contested decision, the Commission found that the exclusivity clause constituted an abuse of a dominant position so far as concerns only the all sites direct partners that had – according to it – typically included in their GSAs all of their websites displaying online search ads. Thus, Article 1(1)(a) and (3) of the operative part of the contested decision concerns only those direct partners that were obliged, by virtue of that clause, to source all or most of their requirements from Google. However, the Commission did not find that the exclusivity clause constituted an abuse of a dominant position so far as concerns those direct partners that had not been considered all sites direct partners.

434    In those conditions, the Commission was limited to finding, in Section 8.3.4.1 of the contested decision, that the exclusivity clause had deterred all sites direct partners from sourcing from Google’s competing intermediaries. Consequently, the question whether other direct partners – that were not all sites direct partners – had been deterred from sourcing from Google’s competing intermediaries is irrelevant.

435    It follows that, as the Commission contends, rightly, the replies of those direct partners that were not all sites direct partners are irrelevant.

(iii) All sites direct partners identified in the contested decision

436    Google claims that the Commission did not establish that the all sites direct partners identified in the contested decision had included all of their websites in their GSAs and that they had consequently sourced all or most of their requirements exclusively from it.

437    Specifically, Google observes that the Commission found, in the contested decision, that all sites direct partners had informed it, in reply to a request for information of 24 February 2017, that they had typically included all of their websites in their GSAs in demonstrating that they were under an exclusive supply obligation. On the one hand, however, the word ‘typically’ implies that those direct partners had been able to exclude some of their websites from the said GSAs. It follows that their reply to that request for information was not sufficiently precise for the Commission to infer that they sourced all or most of their requirements exclusively from Google. On the other hand, the exclusivity clauses in GSAs concluded with [confidential] and [confidential] authorised the display of competing ads on the websites included in those GSAs. Moreover, it is established that 8 other of the 34 all sites direct partners identified in the contested decision had not included all of their websites in their GSAs.

438    Surfboard also claims that the Commission was wrong to find that it had included all of its websites in its GSA. It criticises in that regard the Commission for not having taken into account a letter from its CEO, submitted by Google as an annex to the response to the statement of objections, which indicated that it could use competing online search advertising intermediation services on some of its websites.

439    The Commission contends that Google has not established that all sites direct partners did not source all or most of their requirements exclusively from it.

440    Specifically, the Commission argues that Google has not established that the word ‘typically’ was interpreted differently by direct partners. It is moreover apparent from the line of argument contained in the application that only five of the all sites direct partners could have misinterpreted that word. It states in that respect that the line of argument developed in the reply in relation to other direct partners is belated and, consequently, inadmissible. Last, the Commission considers that, in any event, its approach in identifying all sites direct partners was ‘conservative and favourable to Google’ since the exclusivity clause applied also to 69 other direct partners that had been unable to confirm whether they had typically included all of their websites in their GSAs. As for Surfboard, it notes that that company cites only a single website that was not included in its GSA without however providing any evidence in that regard.

–       Relevance of Google’s line of argument

441    As a preliminary point, it should be recalled that the Court cannot, under any circumstances, substitute its own reasoning for that of the author of the contested act (judgment of 11 November 2021, Autostrada Wielkopolska v Commission and Poland, C‑933/19 P, EU:C:2021:905, paragraph 115). Moreover, the author of that act cannot supplement the statement of reasons for it during the proceedings (see, to that effect, judgment of 15 December 2021, Oltchim v Commission, T‑565/19, EU:T:2021:904, paragraph 275). In the case at hand, however, it is apparent from the contested decision – in particular from recitals 341, 362, 366, 380, 403, 407, 416, 422, 627 and 630, point 1, thereof as well as from Article 1(1)(a) and (3) of its operative part, mentioned in paragraph 433 above – that the Commission found that the exclusivity clause constituted an abuse of a dominant position only in so far as that clause was included in certain GSAs, mentioned in recital 348 of that decision, in which the direct partners concerned had typically included all of their websites, such that the said clause had therefore obliged those direct partners to source all or most of their requirements from Google.

442    Consequently, the circumstance, essentially alleged by the Commission in its written submissions, that that institution could find that the exclusivity clause in other GSAs constituted an abuse of a dominant position, including where the direct partners concerned had not typically included all of their websites in their GSAs, is not such as to refute Google and Surfboard’s line of argument by which those companies criticise the Commission for not having established that the all sites direct partners identified in the contested decision had included all of their websites in their GSAs and that they had sourced at a minimum most of their requirements from Google. Furthermore, it is appropriate to recall that the Commission contends, moreover, in its written submissions, that only the replies of all sites direct partners were relevant to assessing the deterrent effect of the exclusivity clause (see paragraph 435 above), thereby confirming the effectiveness of Google and Surfboard’s line of argument.

–       Reliability of the replies to the request for information of 24 February 2017

443    It should be noted that, where a direct partner had included all of its websites displaying online search ads in its GSA, it necessarily had to source all of its requirements in terms of online search advertising intermediation services exclusively from Google for the duration prescribed by that GSA.

444    However, where a direct partner had not included all of its websites in its GSA, the Commission was required, according to the wording of the contested decision, to demonstrate that those which were included in it represented at a minimum most of the requirements in terms of online search advertising intermediation services of that direct partner. Specifically, it is apparent from recitals 386, 389 and 390 of the contested decision that the volume of online queries, traffic and the revenues generated by the websites were factors relevant to determining the extent of those requirements covered by a GSA.

445    In that regard, it should be noted that the Commission relied primarily on the replies to a question contained in a request for information of 24 February 2017 to identify the direct partners that had sourced all or most of their requirements in terms of online search advertising intermediation services exclusively from Google. That question was worded in the following manner: ‘Does your company typically include all its websites displaying text based search advertisements in the AFS Direct Agreements with Google for the provision of the AFS service?’ The direct partners were requested to answer ‘yes’ or ‘no’. It was also specified that the answer had to take account of the entirety of the undertaking to which that company belonged, including any umbrella companies and subsidiaries.

446    In the first place, it is appropriate to state, as Google does, that the word ‘typically’ meant that direct partners could answer ‘yes’ to the Commission’s question, including where certain of their websites were not included in their GSAs. Thus, that question certainly did not enable the Commission to identify the direct partners that had sourced all of their requirements in terms of online search advertising intermediation services exclusively from Google. However, it is important to recall that the Commission only sought to identify the direct partners that, at a minimum, sourced most of their requirements in terms of its services. It follows that it did not have to demonstrate that all sites direct partners had, in all cases, included all of their websites in their GSAs. Moreover, Google does not explain why it believes that the Commission should have demonstrated that all of the websites of each of those direct partners had to be included in their GSAs.

447    In the second place, Google admittedly argues that the Commission had not defined the word ‘typically’ in the request for information of 24 February 2017. It infers from this that that word could be interpreted differently by direct partners.

448    However, first, while it is true that the word ‘typically’ could be open to two interpretations, that did not prevent the Commission from relying on the replies of direct partners in order to identify those which sourced at a minimum most of their requirements exclusively from Google.

449    The word ‘typically’ was liable to be understood by direct partners as pertaining not only to the number of websites included in GSAs (compared to the number of websites that were excluded from them), but also to the relative ‘importance’ of those websites, in terms of traffic, of the volume of online queries and of sales generated. Thus, for a direct partner to have answered ‘yes’ to the question of whether it had typically included all of its websites in one of its GSAs meant either that the entirety of that direct partner’s websites were included in that GSA or that any websites that were not included in it generated only limited or even negligible traffic, volume of online queries and sales.

450    Second, it should be noted that Google and Surfboard do indeed claim that 11 of the 34 all sites direct partners identified in the contested decision had not included all of their websites in their GSAs and that they had not sourced at a minimum most of their requirements exclusively from Google. Those all sites direct partners are, specifically, Surfboard, [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential] and [confidential].

451    However, on the one hand, it should be noted that, as follows from Annex A.37 to the application and from Annex B.11 to the defence, the [confidential] group (to which belonged [confidential] and [confidential]), the [confidential] group (to which belonged [confidential]) and the [confidential] group had answered ‘no’ to the question of whether the undertaking to which they belonged typically included all of their websites in their GSAs. Consequently, the fact – assuming it were established – that the direct partners concerned had not included all of their websites in their GSAs cannot be an indication that the said question did not enable identification of the direct partners that had sourced at a minimum most of their requirements exclusively from Google.

452    On the other hand, it is necessary to note that, while Google criticises the Commission for not having established that [confidential], [confidential], [confidential], [confidential] and [confidential] had sourced at a minimum most of their requirements exclusively from Google, it does not indicate that those direct partners – or the groups to which they belonged – had answered ‘yes’ to the question of whether they typically included all of their websites in their GSAs.

453    In those conditions, it should be pointed out that Google and Surfboard are limited to arguing that only 2 of the 34 all sites direct partners identified in the contested decision, namely Surfboard and [confidential], or the groups to which they belonged, had answered ‘yes’ to the question of whether they typically included all of their websites in their GSAs, when those direct partners had not included all of their websites in their GSAs. Even supposing, however, that those direct partners had not included all of their websites in their GSAs and that the Commission consequently had had to ascertain the extent of the requirements of the said direct partners in terms of online search advertising intermediation services covered by those GSAs, those two examples cannot be considered, in any event, sufficient to demonstrate that the word ‘typically’ in itself posed such a difficulty of interpretation that it did not allow the existence of an exclusive supply obligation towards any of those all sites direct partners to be established.

454    Accordingly, without it being necessary to rule on the admissibility, disputed by the Commission, of the line of argument of Google contained in the reply, it must be held that Google is not entitled to claim that the use of the word ‘typically’ in the request for information of 24 February 2017 alone could lead the Commission to consider, wrongly, that all sites direct partners were under an exclusive supply obligation for all or most of their requirements.

455    In those conditions, without it being necessary to rule on the question of whether the 11 direct partners mentioned in paragraph 450 above had sourced at a minimum most of their requirements exclusively from Google, it is appropriate to hold that the 23 other direct partners mentioned in recital 348 of the contested decision must be regarded as being all sites direct partners which had sourced at a minimum most of their requirements exclusively from Google.

(iv) Replies of the all sites direct partners mentioned in recitals 367 and 368 of the contested decision

456    As follows from paragraph 421 above, the Commission referred, in recitals 367 and 368 of the contested decision, to eight replies reflecting the position of 10 of the 34 all sites direct partners that it had identified. Those are the replies of [confidential], the [confidential] group, [confidential], the [confidential] group, the [confidential] group, the [confidential] group, [confidential] and the [confidential] group.

457    Google challenges the tenor of five of the eight replies quoted in recitals 367 and 368 of the contested decision, namely those of [confidential], the [confidential] group, [confidential], the [confidential] group and the [confidential] group, but does not challenge that of the replies of the [confidential] group, [confidential] and the [confidential] group.

458    As a preliminary point, it should be emphasised that, as is apparent from the wording of recitals 367 and 368 of the contested decision, the Commission did not intend to compile an exhaustive list of all the replies that it received during the administrative procedure, but it is limited to giving only examples of replies corroborating the fact that, on the one hand, all sites direct partners could be deterred from sourcing at least part of their requirements from Google’s competing intermediaries due to the exclusivity clause and that, on the other hand, that clause was capable of preventing the said direct partners from evaluating the commercial interest in sourcing from such intermediaries.

–       Replies of [confidential]

459    In recital 367 of the contested decision, the Commission quoted the following extract of a reply of [confidential] to the request for information of 22 December 2010: ‘[the] exclusivity clauses prevented [confidential] from using providers of sponsored links’. It is apparent from recital 348 of the contested decision that [confidential] was one of the all sites direct partners identified by the Commission.

460    Google submits that, in reply to Question 14 of the request for information of 18 March 2016, which corresponded, in essence, to Question 5.2 of the request for information of 22 December 2010, as regards the period between 2011 and 2015, [confidential] had stated the following: ‘Except a few experimentations, we only work with Google, because there is no viable alternative that would allow us to generate the same level of revenue we do with Google.’

461    However, the fact that, as is apparent from paragraph 460 above, [confidential] indicated that, in 2016, it worked only with Google on account of the lack of viable alternatives allowing it to generate the same level of revenue as that generated using Google’s services does not suffice to call into question the assertion made by [confidential] in 2010 and quoted in recital 367, point 1, of the contested decision, according to which the exclusivity clause had prevented it from using other ‘providers of sponsored links’.

462    On the one hand, the reply cited by Google seems to pertain specifically to the period between 2011 and 2015, whereas that relied on by the Commission concerned the period before 2011. As has essentially been recalled in paragraph 402 above, however, the Commission noted, in recital 276 of the contested decision, that Google’s share on the market for online search advertising intermediation had increased consistently between 2006 and 2016, such that, in 2016, almost none of Google’s competitors on that market remained, it being specified, moreover, that the effects of scale and network effects rendered the emergence of new competitors difficult. Thus, the fact that [confidential] noted, in 2016, the absence of a ‘viable alternative’ allowing it to generate the same level of revenue as that generated using Google’s services between 2011 and 2015 actually reflects the evolution of Google’s market share, which shows that, during that period, there were almost no viable alternatives left on the market. However, the said reply does not mean that the exclusivity clause had not deterred it from sourcing from a competitor between 2006 and 2010, as is explicitly apparent from its reply to the request for information of 22 December 2010 quoted in recital 367 of the contested decision.

463    On the other hand, even assuming that Google’s services were of superior quality and had allowed [confidential] to generate a higher level of revenue than that generated using the services of Google’s other competing intermediaries, such a circumstance does not necessarily mean that [confidential] had no commercial interest in sourcing, at least part of its requirements, from such intermediaries.

464    In that regard, first, as the Commission observed in recital 377 of the contested decision, the fact that Google concluded GSAs containing an exclusivity clause is an indication that, notwithstanding the supposed superior quality of its services, Google considered that those partners would have a commercial interest in sourcing online search ads from other intermediaries. After all, had Google considered, as it argues before the Court, that, even in the absence of the exclusivity clause, all sites direct partners would still have chosen AFS on account of its alleged superior quality, it would not have needed to include such a clause in GSAs.

465    Second and in any event, the quality of a service is but one factor among others that an economic operator takes into consideration when it decides from where to source its supplies. Other important factors include, for example, the price of that service or, in the case of the market for online search advertising intermediation, the share of revenues allocated to the publisher of the said service. Thus, the mere circumstance that the quality of a service is supposedly superior to that offered by a competitor does not necessarily suffice to strip an exclusivity clause, such as that at issue in the case at hand, of all of its deterrent effect vis-à-vis those economic operators.

466    In addition, the better monetisation of online search ads enabled by AFS is, even if only in part, an inherent consequence of the network effects, described in paragraph 403 above, which characterise Google’s dominant position on the market for online search advertising intermediation and allowed AFS to increase exponentially the relevance of the ads shown in response to users’ online queries. That better monetisation is thus, at least in part, the result of the said network effects and not necessarily that of the supposed superior quality of the services offered by Google.

467    In those conditions, it is appropriate to find that the Commission considered, rightly, that the reply of [confidential] was capable of corroborating the fact that the exclusivity clause could have deterred all sites direct partners from sourcing at least part of their requirements from Google’s competing intermediaries.

–       Replies of the [confidential] group

468    In recitals 367 and 368 of the contested decision, the Commission quoted the following extracts from the replies of the [confidential] group to Questions 5.2.c and 5.2.e of the request for information of 22 December 2010 relating respectively to the reasons why that group had chosen to partner with only one intermediary at a time and to the costs necessary for changing intermediaries: ‘We use one intermediary for syndicated search ads because until recently, our contract with Google did not allow us to utilize another vendor and so we were prevented from engaging another vendor’ and [confidential].

469    In that regard, Google claims that, in reply to Question 5.2.d of the request for information of 22 December 2010, the [confidential] group had indicated that it would consider switching advertising intermediary if it identified another providing ‘better financial performance’, either directly (for example by a higher revenue per click), or indirectly (for example by allowing more flexibility to optimise results in a way that generates higher revenue per click, and by providing revenue estimates on a keyword basis).

470    The Commission disputes Google’s line of argument.

471    It should be noted that it is clear from the extracts of the replies of the [confidential] group, quoted in paragraph 468 above, that that group had chosen to source from only one intermediary at a time on account of the exclusivity clause. That conclusion is not called into question by the assertion in paragraph 469 above, which concerns a future possibility of a hypothetical nature.

472    In those conditions, and as Google moreover acknowledged in its response to a question from the Court during the hearing, it is appropriate to find that the Commission was right to consider that the replies of the [confidential] group were capable of corroborating the fact that the exclusivity clause could have deterred all sites direct partners from sourcing at least part of their requirements from a competing intermediary.

–       Replies of [confidential]

473    In recital 367 of the contested decision, the Commission quoted the following extract of a reply of [confidential] to Question 8.5 of the request for information of 22 December 2010 on whether the exclusivity clause had had an impact on its advertising strategy: ‘The exclusivity clauses in question have had a significant impact on our advertising strategy, particularly when we first contemplated adding third party text advertising to our websites … Since Google would not permit us to work with both companies, we maximized our revenue by signing with Google on an exclusive basis and foregoing any opportunity to work with Yahoo or other text advertising service.’

474    In that regard, Google argues that [confidential] also replied to Question 8.5 of the request for information of 22 December 2010 in the following manner: ‘Google traditionally has had higher [cost-per-click] rates than competitors like Yahoo. Despite Yahoo being willing to share a larger percentage of [cost-per-click] revenue with [[confidential]], the overall return from the Google product was still higher than from Yahoo.’

475    As has been noted in paragraph 473 above, it is apparent from the reply of [confidential] to Question 8.5 of the request for information of 22 December 2010 that the exclusivity clause had a ‘significant impact’ on its advertising strategy, on the ground that that clause had prevented it from procuring online search advertising intermediation services from Google and from one or more of its competitors at the same time. It may thus be inferred that, particularly in the beginning, when it defined its advertising strategy, [confidential] wished to source at least part of its requirements from one of Google’s competitors. It is true that it is also apparent from that reply that [confidential] considered that AFS generated more revenues than Yahoo!’s services. However, in line with what has been found in paragraphs 464 to 466 above, such a circumstance does not suffice to demonstrate that [confidential] had no commercial interest in sourcing, at least part of its requirements, from Yahoo! or another of Google’s competitors.

476    Consequently, it is appropriate to find that the Commission was right to consider that the reply of [confidential] was capable of corroborating the fact that the exclusivity clause could have deterred all sites direct partners from sourcing at least part of their requirements from Google’s other competing intermediaries.

–       Replies of the [confidential] group

477    In recital 367 of the contested decision, the Commission quoted the following extract of the reply of the [confidential] group to Question 8.5 of the request for information of 22 December 2010: ‘We would consider displaying competing search ads on our webpages and … would consider partnering with Yahoo, Bing and/or other suppliers of search ads for that purpose.’

478    In that regard, on the one hand, Google claims that, in reply to the question as to the basis on which it chose the type of ads to be placed on the various advertising spaces of its webpages and to Question 5.2.d, the [confidential] group had indicated that it took into account (i) revenue maximisation, (ii) user response, (iii) contractual obligations, and (iv) the nature of its business, namely the fact that the ads that it displayed were predominantly online search ads since it operated a search engine.

479    On the other hand, Google notes that, in reply to Question 8.6 of the request for information of 22 December 2010, relating to the reasons why the [confidential] group had accepted the exclusivity clause, that group had indicated that it attracted more advertisers than any of its competitors and that, therefore, its online search ads generated higher revenues overall than those of its competitors.

480    The Commission disputes Google’s line of argument.

481    As has been noted in paragraph 477 above, it is apparent from the reply of the [confidential] group to Question 8.5 of the request for information of 22 December 2010 that it ‘would consider’ entering into a partnership with one of Google’s competitors were it not for the exclusivity clause. In that regard, it should be noted that that reply is consistent with the reply of that same group to Question 5.2.d, in which it had indicated that it took into account inter alia its contractual obligations in choosing whether or not to change intermediaries. It is true that the [confidential] group appeared to consider, in its reply to Question 8.6, that Google offered a service superior to that of its competitors. However, in line with what has been found in paragraphs 464 to 466 above, such a circumstance is not capable of demonstrating that the [confidential] group had no commercial interest in sourcing, at least part of its requirements, from another intermediary.

482    In those conditions, it is appropriate to find that the Commission was right to consider that the reply of the [confidential] group was capable of corroborating the fact that the exclusivity clause could have deterred all sites direct partners from sourcing at least part of their requirements from Google’s competing intermediaries.

–       Replies of the [confidential] group

483    In recital 367 of the contested decision, the Commission quoted the following extract of a reply of the [confidential] group to Question 8.9 of the request for information of 22 December 2010, on whether there were any investments from that group or from Google that would not have taken place in the absence of the exclusivity clause: ‘[That clause] has meant that developments and partnerships with other market players needed to be considered thoroughly and possibly even delayed or rejected.’

484    In that regard, Google argues that, in reply to Question 8.6 of the request for information of 22 December 2010, on the reasons why the [confidential] group had accepted the exclusivity clause, it had indicated, first, that, in Finland, ‘the advantage was that the Google ads [had] provided good revenue for an ad space on a certain web site’, next, that, in the Netherlands, ‘Google provided advanced technological and profitable advertising solutions, which were proven successful’ and that ‘the gross revenue [had] increased substantially since 2005, because the AdSense service [had been] optimised by Google’ and, last, that, in Hungary, ‘the analysis [had] showed that Google’s service [had been] the only of a kind available as to profitability and technology in Hungarian market’.

485    The Commission contends that Google’s line of argument concerning Finland and Hungary is irrelevant in so far as only the entities of the [confidential] group established in the Netherlands, namely [confidential] and [confidential], are all sites direct partners. As for Google’s line of argument relating to the Netherlands, it argues that, as it had noted in recital 367, point 5, of the contested decision, the [confidential] group had also indicated that the exclusivity clause ‘meant that developments and partnerships with other market players needed to be considered thoroughly and possibly even delayed or rejected’.

486    It should be noted that, in recital 355, point 6, of the contested decision, as clarified by a response to a measure of organisation of procedure, the Commission considered that only [confidential] and [confidential] figured among the all sites direct partners. It follows that the other entities of the [confidential] group, including those situated in Finland and Hungary, are not among those direct partners.

487    However, first, it should be recalled, as has been mentioned in paragraph 67 above and as follows inter alia from recitals 338, 341 and 347 to 349 of the contested decision, the Commission found that all sites direct partners were those that had typically included, at a given moment, all of their websites displaying online search ads in at least one of their GSAs. It follows that those direct partners were, consequently, purchasers, within the meaning of the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), which were under an obligation to source all or most of their requirements from Google. It should be pointed out, however, that the concept of ‘purchaser’, within the meaning of that case-law, corresponds to an undertaking, taken as a whole, which in the case at hand would be the [confidential] group and not merely one of its subsidiaries. In the context of competition law, the concept of undertaking covers any entity engaged in an economic activity, irrespective of its legal status (judgments of 23 April 1991, Höfner and Elser, C‑41/90, EU:C:1991:161, paragraph 21, and of 11 June 2020, Commission and Slovak Republic v Dôvera zdravotná poist’ovňa, C‑262/18 P and C‑271/18 P, EU:C:2020:450, paragraph 28). Therefore, the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), on which the contested decision is based, cannot justify not taking into consideration the replies of the other legal entities belonging to the same group as the all sites direct partners.

488    Moreover, the request for information of 24 February 2017, by which the Commission had identified the all sites direct partners, asking them whether or not they had typically included all of their websites in their GSAs, specifically stated that the expected reply had to take account of the entirety of the undertaking concerned, whether it be the parent company or subsidiaries.

489    Second, in recitals 218 to 221 of the contested decision, the Commission defined the market for online search advertising intermediation as covering the entirety of the EEA. It follows that the Commission could not merely take into account the replies of the [confidential] group relating to certain of its legal entities established in the Netherlands to the exclusion of those established in other Member States.

490    Third, it would appear from recitals 367 and 368 of the contested decision, as well as from the questions mentioned in footnotes 493 to 502 to that decision, that the Commission assessed the deterrent effect of the exclusivity clause on the basis of the replies of the groups to which the all sites direct partners belonged, and not on the basis of the specific answers of those direct partners as identified in recital 348 of that decision. Consequently, the Commission could not, as regards the [confidential] group in particular, deem the replies of certain of that group’s legal entities irrelevant.

491    In those conditions, it is necessary to find that the Commission had to take into account all of the replies of the groups that had at least one related entity among the all sites direct partners.

492    Consequently, contrary to what the Commission argues, it must be held that the replies of the [confidential] group relating to Finland and Hungary could not possibly have been considered irrelevant.

493    However, it must be noted that the [confidential] group explicitly stated that the exclusivity clause could delay a partnership with one of Google’s competitors or lead it to refuse such a partnership. Consequently, while the [confidential] group indicated that it had accepted that clause on the ground that the services offered by Google were superior to those of its competitors, in particular as regards Hungary, it must be noted, in line with what has been found in paragraphs 464 to 466 above, that such a circumstance is not sufficient to demonstrate that the said group had no commercial interest in sourcing, at least part of its requirements, from another intermediary.

494    Consequently, it is appropriate to find that the Commission was right to consider that the reply of the [confidential] group was capable of corroborating the fact that the exclusivity clause could have deterred all sites direct partners from sourcing at least part of their requirements from Google’s competing intermediaries.

–       Replies of the [confidential] group, [confidential] and the [confidential] group

495    It is apparent from recital 367 of the contested decision that, according to [confidential], absent the exclusivity clause, it would have experimented with other competing intermediaries of Google, such as Yahoo! and Microsoft. In addition, according to [confidential], were it not for the said clause, it would have considered sourcing from Google’s competing intermediaries, provided that that would enhance its revenues. It is apparent from recital 368 of the contested decision that, according to the [confidential] group, the exclusivity clause prevented it from ‘begin[ning] testing other providers’ and ‘upgrading [its] system’ in order to be able to work with other providers.

496    As has been noted in paragraph 457 above, Google does not challenge the tenor of those replies.

(v)    Other replies of the all sites direct partners cited by Google

497    Google cites the replies of all sites direct partners or of the groups to which those direct partners belonged, including, in particular, [confidential], [confidential] (which together constitute one of the all sites direct partners identified by the Commission), [confidential], [confidential], [confidential], the [confidential] group, to which belonged [confidential] (which together constitute one of the all sites direct partners identified by the Commission), the [confidential] group, to which [confidential] belonged, and the [confidential] group, to which belonged [confidential] and [confidential] (which constitute separate all sites direct partners identified by the Commission). It is apparent from those replies that the direct partners had chosen AFS on account of its superior qualities, such that the exclusivity clause had not deterred them from sourcing from other competing intermediaries of Google.

–       Replies of [confidential] and [confidential], [confidential], [confidential], the [confidential] group and the [confidential] group

498    First of all, Google argues that, in reply to Question 5.2.d of the request for information of 22 December 2010, [confidential], [confidential], [confidential] and [confidential] had each indicated to the Commission that their choice of intermediary depended on the revenues it generated. It also notes that [confidential] and [confidential] moreover stated that they would consider switching intermediaries if a different intermediary enabled them to generate more revenues than Google.

499    Next, Google notes that the [confidential] group had indicated, in the context of the request for information of 18 March 2016, first, in reply to the question of why it had chosen Google as intermediary, that it ‘ha[d] been chosen as sole intermediary as it [wa]s the market leader for search ads with the best monetisation potential and best revenue perspectives’, and, second, in reply to the question of what were the ‘material differences’ between the services offered by the various intermediaries, that it ‘[did] not see that there [was] another potential partner [other than Google] who could provide [it] with better search ad products and generate more revenue from search advertising’. It also stated that, if another service offering a similar monetisation potential appeared, the costs of switching should be low, or even nil.

500    Last, Google notes that the [confidential] group had indicated, in reply to Question 9.7 of the request for information of 31 July 2015 on the reasons for which it had accepted the exclusivity clause, that it enabled it to change intermediary for mobile queries, for which alternatives to Google could be found in France in 2014, and that, ‘Google being a leader in its domain, [that clause] allow[ed it] the best monetization possible for search on desktop at least’.

501    In that regard, it is apparent from those replies, in essence, that those all sites direct partners considered Google the market leader and that AFS offered them better monetisation than that of the services of Google’s competitors. It is sufficient to hold, however, in line with what has been noted in paragraphs 464 to 466 above, that such a circumstance, which can be explained at least in part by the network effects characterising Google’s dominant position, does not suffice to demonstrate that those direct partners had no commercial interest in sourcing, at least part of their requirements, from competing intermediaries.

502    In those conditions, those replies cannot be regarded as capable of calling into question the Commission’s conclusion that the exclusivity clause could have deterred all sites direct partners from sourcing at least part of their requirements from Google’s competing intermediaries.

–       Replies of [confidential] and the [confidential] group

503    In the first place, Google argues that [confidential] had stated, in reply to Question 5.2.a (‘why have you chosen … the intermediary you are currently working with’) of the request for information of 22 December 2010, that ‘Google [had] initially [been] selected … given its market coverage and ability to generate revenue’, that ‘Yahoo [had been] selected as the intermediary in Sept[ember] 2008 because Yahoo offered the ability to syndicate their Sponsored Links, along with [confidential] sold Sponsored Links, to third party travel sites’, that ‘Yahoo also promised monetization similar to Google [and that] [confidential] [had] switched back to Google in Jan[uary] 2009 due to Yahoo’s inability to monetize at the same level as Google’. In addition, Google notes that [confidential] had stated, in reply to Question 5.2.d of that request for information, that it ‘would opt to switch or more likely add another intermediary if it could be certain it would improve its product or improve its monetization’. Last, it states that, in reply to Question 8.5 of the said request for information, [confidential] had indicated that its ‘advertising strategy ha[d] not been affected by the exclusivity clauses’.

504    The Commission submits that the replies of [confidential] corroborate the fact that the exclusivity clause had deterred all sites direct partners from sourcing from Google’s competing intermediaries. It quotes, in that regard, without providing further explanation, the reply of [confidential] to Question 5.2.d, cited by Google and mentioned in paragraph 503 above, as well as that to Question 8.6, according to which ‘[confidential] [had] initially accepted the exclusivity clauses because Google required them as a condition of the agreement’.

505    In that regard, it is apparent from the replies of [confidential] quoted in paragraph 503 above that, unlike the all sites direct partners, [confidential] clearly stated that its ‘advertising strategy had not been affected by the exclusivity clauses’.

506    In the second place, it is appropriate to note, as Google does, that the [confidential] group had indicated, in reply to the request for information of 22 December 2010, that its ‘strategy [had] not [been] impacted’ by the exclusivity clause and that it ‘would not integrate ads from more than one … provider [in the absence of that clause] and [that it was] not restricted from displaying non-search ads on its websites’.

507    In those conditions, the replies of [confidential] and the [confidential] group should be deemed such as to corroborate Google’s argument that the exclusivity clause had not deterred those all sites direct partners from sourcing at least part of their requirements from another intermediary, which the Commission does not dispute.

(vi) Surfboard’s letter

508    It is important to note that Surfboard’s CEO had explained, in a letter nominally addressed to the Commission, but provided to it by Google as an annex to the response to the statement of objections, that the exclusivity clause had not had any effect on Surfboard’s conduct. He had indicated in that regard that the GSA at issue containing that clause applied to Surfboard’s ‘main’ websites, on the ground that the revenues generated by AFS were ‘substantially higher’ than those generated by Yahoo!’s services. However, he had also stated that other websites, including the website ‘www.ixquick.eu’, had not been included in that GSA.

509    In recital 370 of the contested decision, the Commission considered that the probative value of the letter of Surfboard’s CEO was limited for two reasons. On the one hand, Surfboard had previously indicated, in reply to a request for information, that it had typically included all of its websites in its GSAs. On the other hand, the Commission did not know how Google had obtained that letter.

510    Surfboard criticises the Commission for not having taken into account the letter of its CEO that had been provided to it as an annex to the response to the statement of objections. That latter explained that Surfboard had used Yahoo!’s services on one of its websites. Moreover, it claims that it chose AFS on its main websites on the basis of technical and financial considerations.

511    The Commission contends that Surfboard refers only to one website that was not included in its GSA. In addition, it maintains that the letter of Surfboard’s CEO, that Google had provided to it, was ambiguous. Furthermore, it argues that the probative value of that letter is more limited than that of Surfboard’s reply to a request for information by which that company informed it that it typically included all of its websites in the GSA at issue. On the one hand, the context in which Google obtained the said letter is unknown. On the other hand, Surfboard could provide, in that letter, incorrect or misleading information without the risk of being imposed a fine pursuant to Article 23(1)(a) of Regulation No 1/2003.

512    In that regard, it should be recalled that the guarantees afforded by EU law in administrative proceedings include, in particular, the principle of sound administration, which is enshrined in Article 41 of the Charter of Fundamental Rights of the European Union, which entails the duty of the competent institution to examine carefully and impartially all the relevant aspects of the individual case (judgment of 27 September 2012, Shell Petroleum and Others v Commission, T‑343/06, EU:T:2012:478, paragraph 170).

513    First, it should be noted that, unlike the Commission, Google did not enjoy the investigative and sanctioning powers instituted by Regulation No 1/2003, meaning that it necessarily had to rely on the voluntary cooperation of direct partners to collect information relating inter alia to the websites that they had not included in their GSAs in order to be able to defend itself in the present case. However, this does not mean that the information collected by Google was necessarily irrelevant, on the ground that its probative value was more limited.

514    Moreover, it should be noted that the Commission merely indicates that it did not know how Google had obtained the letter of Surfboard’s CEO that was addressed to it, but that it does not call that letter’s authenticity into question or argue that it is devoid of all probative value. Furthermore, it should be noted that the Commission could, under Article 18 of Regulation No 1/2003, request further information from Surfboard directly if it believed that the said letter could not be reconciled with that company’s reply to a previous request for information.

515    Second, it should be noted that, contrary to what the Commission contends, the letter of Surfboard’s CEO is not ambiguous. It is clear from it that Surfboard had not included at least one of its websites in the GSA at issue.

516    Third, it should be noted, as Surfboard does, that the fact that that company indicated, in reply to a request for information of the Commission, that it had typically included all of its websites in the GSA at issue did not exclude the fact that certain of its websites might not have been included in it. Thus, it must be held that the letter from Surfboard’s CEO and the reply of that company to the Commission’s request for information were not contradictory.

517    In those conditions, the Commission should have found that Surfboard’s letter, in which that company stated that the exclusivity clause had not had any effect on its conduct, was such as to call into question the fact that the said clause had deterred all sites direct partners from sourcing at least part of their requirements from another intermediary.

(vii) Conclusion on the replies of direct partners to the Commission’s various requests for information and Surfboard’s letter

518    Having regard to the foregoing, it is appropriate to find that the Commission could rightly take into consideration the examples of replies of direct partners set out in recitals 367 and 368 of the contested decision as elements capable of corroborating its assessment that the exclusivity clause in GSAs concluded with all sites direct partners could have deterred them from sourcing at least part of their requirements from Google’s competing intermediaries. By contrast, as has been found in paragraphs 503 to 517 above, the replies of certain of the other all sites direct partners or of the groups to which they belonged, as well as Surfboard’s letter, were not capable of corroborating such an assessment.

519    Moreover, it should be found that, while Google disputes that certain of the replies quoted in recitals 367 and 368 of the contested decision, namely those of the [confidential], [confidential], [confidential] and [confidential] groups, reflecting the position of direct partners that had sourced at a minimum most of their requirements exclusively from Google pursuant to the exclusivity clause, it follows from paragraph 455 above that the replies of the other direct partners or of the groups to which they belonged must be considered to be coming from all sites direct partners. In particular, as regards the [confidential] group, it should be noted that Google disputes only that one of the companies belonging to that group, namely [confidential], was an all sites direct partner, but it does not dispute that another of the companies belonging to the same group, namely [confidential], had sourced at a minimum most of its requirements exclusively from Google. In those conditions, and even assuming that Google’s line of argument could succeed, that is to say, in the scenario most favourable to Google, it would not affect the evidence presented by the Commission and recalled in paragraph 457, so far as concerns [confidential] and the [confidential] group, and in paragraphs 459 to 467 and 473 to 482 above, which tend to corroborate, as one relevant element among others, that the exclusivity clause was capable of deterring all sites direct partners from sourcing from Google’s competing intermediaries.

(3)    Investments made by Yahoo!

520    In the contested decision, the Commission found that Yahoo! had made substantial capital investments in its general search service between 2006 and 2015. It found, in recital 402 of the said decision, that a ‘2006 internal Google document’ confirmed that the level of those investments was ‘similar’ to that of Google.

521    Google claims that, as it had demonstrated during the administrative procedure, AFS was a service superior to that of its competitors, because those competitors had failed to invest adequately in their offerings. Inter alia Yahoo! and Microsoft did not, in essence, make ‘effective investments’ in technology development and localisation. Moreover, it notes that the amount of Yahoo!’s investments, presented in the contested decision, took account of the purchase and sale of real estate. It infers from this that that amount revealed nothing about the scale of the investments made by Yahoo! specifically in its online search service.

522    The Commission contends that it is apparent from recitals 401 and 402 of the contested decision, the content of which is not disputed, that Yahoo! had reported substantial capital investments in its general online search service between 2006 and 2015 which were comparable to Google’s. It adds that the claim made by Google in the reply according to which those investments included the purchase and sale of real estate is belated and, therefore, inadmissible.

523    In that regard, it should be noted that Google had described to the Commission, in its response to the statement of objections, the reasons for which it believed that AFS was regarded by publishers as being a better service than that of Yahoo! and Microsoft. First, it had noted that the precursor technology, developed by Overture and acquired by Yahoo! in 2003, had not been designed to be used on a global scale and that it was significantly slower than Google’s in order to allow for human review of each ad displayed on a website. Second, it had experienced problems integrating the teams that had participated in the development of the technology acquired by Yahoo!, leading to the departure of several engineering chiefs. Third, it had noted that Yahoo! had started to take account of CTRs, to determine the ads to show in response to an online query, only from 2007, whereas it is apparent from exhibit 36 of Annex C.1 to the reply, to which the response to the statement of objections referred, that Google had developed that feature since 2002. Moreover, it had noted that, despite that change, Yahoo! had publicly acknowledged in 2008 that AFS still offered better monetisation, the gap between AFS and Yahoo!’s service having narrowed only by 30%, according to that latter company. Fourth, Google had noted that the implementation of the partnership between Yahoo! and Microsoft had been slow. It had found, after all, that that partnership had been concluded in 2009, but that [confidential]. Fifth, it had noted that the said partnership had not achieved its objectives. In that regard, it had noted that Yahoo! had publicly recognised the ‘technical limitations’ of Microsoft’s platform in 2011, that Yahoo! and Microsoft’s teams responsible for publishers in the EEA amounted to fewer than 20 people each, whereas Google’s teams reached approximately 300 people and Yahoo! and Microsoft had been slow to adapt their online search advertising services to each Member State to take into account user location.

524    The Commission did not dispute Google’s claims. It was limited, in recitals 401 and 402 of the contested decision, on the one hand, to determining the yearly amount of Yahoo!’s capital investments in its general search service between 2006 and 2016 and, on the other hand, to quoting extracts from a 2006 Google internal document to find that the level of investments of Yahoo! and Google was comparable.

525    However, assuming, as the Commission contends, that the yearly amount of Yahoo!’s capital investments in its general search service between 2006 and 2016 was relevant to assessing the scale of the investments made by that company in its online search advertising intermediation service, it is appropriate to note that it cannot follow solely from the 2006 Google internal document, quoted in the contested decision, that the investment amounts of Google and Yahoo! had been similar for the entirety of the period between 1 January 2006 and 31 March 2016. In particular, even though the said document already referred to the ‘merger’ between Microsoft and Yahoo! and to the possible effect of that merger on investment, it should be borne in mind that the author of that document could not predict the subsequent events cited by Google and mentioned in paragraph 523 above.

526    Consequently, without it being necessary to rule on the admissibility of Google’s line of argument aimed at calling into question the exact amount of the investments made by Yahoo!, it must be noted that it is not established that those investments had been similar to those made by Google.

527    That being so, it is appropriate to note that, irrespective of whether and to what extent the amount of the investments made by Yahoo! was comparable to that of the investments made by Google, it is apparent from recital 401 of the contested decision that the amount of Yahoo!’s investments was, in any event, substantial. It is for that reason that the Commission was right to reject, in the contested decision, Google’s argument that it was due to Yahoo!’s inadequate investments, and not the exclusivity clause, that the latter company had not been able to access a significant part of the market.

528    Moreover, it is apparent from Table 8 of the contested decision that Yahoo!’s market share on the online search advertising intermediation market had dropped significantly between 2006 and 2008, the period during which the exclusivity clause featured in the template GSA.

529    Consequently, the circumstance, assuming it were established, that Yahoo!’s investments had not enabled that company to develop an online search advertising intermediation service as efficient as that of AFS does not suffice to demonstrate that the exclusivity clause in GSAs concluded with all sites direct partners had not had a deterrent effect on them.

(4)    Preference of publishers for procuring online search ads from one intermediary at a time

530    In the contested decision, on the one hand, the Commission explained that, for all sites direct partners, online contracts did not constitute an alternative to GSAs. Relying, in that regard, on the hearings before the FTC, on 2 and 3 May 2012, of [confidential], then [confidential] at Google, respectively, as well as on Google’s internal guidelines, it found that online contracts were standardised contracts which did not allow those direct partners’ ‘specific needs’ to be met.

531    On the other hand, the Commission noted that the analysis of the conduct of online partners was irrelevant because their needs and those of direct partners were different. It inferred that the study produced by Google during the administrative procedure, showing that online partners used AFS almost exclusively on the websites included in their online contracts, despite the absence of the exclusivity clause from those contracts, was irrelevant.

532    Google claims that the exclusivity clause could not produce a foreclosure effect since most direct partners wished to source from only one intermediary at a time. In that regard, first, it relies on the study, mentioned in paragraph 531 above, from which it is apparent that only [confidential]% of the domains corresponding to the websites of online partners generating revenues in the EEA used AFS and Yahoo! or Microsoft’s services simultaneously. Second, it notes that the replies of direct partners to various requests for information of the Commission confirm the fact that direct partners would have chosen to source from a single intermediary absent the said clause.

(i)    Study produced by Google during the administrative procedure

533    Google claims, in essence, that direct partners – including all sites direct partners – had the possibility of concluding online contracts and that they in practice concluded such contracts, such that publishers could be both direct partners and online partners at the same time. It adds that the ‘alleged differences’ between the needs of direct partners and those of online partners, identified in the contested decision, did not affect the choice of the latter to source from one intermediary at a time or from several intermediaries simultaneously. It thus considers that the conduct of online partners was relevant to analysing the conduct that all sites direct partners could have adopted absent the exclusivity clause with regard to the periods in which those direct partners had typically included all of their websites in their GSAs.

534    Surfboard claims that publishers could conclude online contracts in relation to websites which did not require Google to meet specific requirements. It adds that negotiating a GSA allowed it to secure better terms than those offered under an online contract.

535    The Commission disputes Google and Surfboard’s line of argument.

536    In that regard, it is apparent from Annex A.42 to the application that large publishers, certain of which were all sites direct partners identified in the contested decision, had used AFS for certain of their websites and online contracts for others. The Commission does not dispute the content of that annex, as it confirmed during the hearing.

537    However, first, it must be pointed out that no abuse of a dominant position was found, in the contested decision, as far as online partners were concerned.

538    Second, it should be noted that, while Annex A.42 to the application mentions websites, belonging to large publishers, which had been included in online contracts, it contains no information enabling a determination of the proportion of online search advertising revenues generated by those websites compared to those of those same publishers which were included in GSAs. Nor does it specify the duration for which the said websites had been included in online contracts.

539    Third, it follows from recitals 76 and 371, points 1 and 2, of the contested decision, the content of which is not disputed by Google, that GSAs were individually negotiated by each publisher and were, in principle, reserved to the largest publishers capable of generating revenues sufficiently high to justify the costs incurred by Google related to the staff and support services dedicated to direct partners. In addition, Google claims that GSAs offered publishers a more advantageous revenue share than that offered by online contracts, which were standard, non-negotiable contracts. Likewise, Surfboard claims that concluding a GSA had allowed it to secure more favourable terms than those offered by an online contract.

540    In those conditions, on the one hand, although, as Google claims, all publishers could conclude an online contract, Google itself determined which publishers could conclude GSAs. On the other hand, where publishers had the choice of concluding a GSA or an online contract, it was in principle in their interest to conclude a GSA rather than an online contract.

541    It follows that, without further explanations from Google, the examples, mentioned in Annex A.42 to the application, of direct partners that had also concluded, at a given time and for certain of their websites, an online contract, must be regarded as specific examples which do not necessarily reflect the conduct of direct partners as a whole.

542    Consequently, it is appropriate to find that direct partners and online partners constituted, in principle, two distinct categories of publisher, such that the conduct of online partners does not constitute a sufficiently reliable indication for determining whether all sites direct partners would have sourced exclusively from Google had it not been for the exclusivity clause.

543    It follows that the Commission was right to find that the study produced by Google during the administrative procedure was irrelevant.

(ii) Replies of direct partners cited by Google

544    Google claims that many direct partners indicated, in reply to various requests for information of the Commission, that they did not wish to use competing online search advertising intermediation services simultaneously. It states, in that regard, that the Commission has not alleged that the conduct of all sites direct partners was different from that of other direct partners.

545    Surfboard claims that there was no economic incentive to display ads from different intermediaries on the same page and that it had historically preferred to work with only one intermediary at a time. It states, in that regard, that close collaboration between the intermediary, on the one hand, and Surfboard’s editorial office and product manager, on the other hand, was required and that the simultaneous use of different services was such as to increase the risk of displaying redundant or poor-quality ads. Moreover, it adds that most intermediaries require exclusivity in relation to the inventory they acquire in order to meet advertisers’ requirements. Thus, the exclusivity clause results not from Google’s market power, but from the desire of advertisers to secure high-quality inventory.

546    The Commission disputes Google and Surfboard’s line of argument.

547    In the first place, it should be recalled that the Commission found that the exclusivity clause had a deterrent effect only vis-à-vis all sites direct partners. It follows that only the replies of those direct partners are relevant to determining whether those same direct partners would have sourced from Google’s competing intermediaries in the course of the period during which they had typically included all of their websites in their GSAs. The circumstance, invoked by Google, according to which the Commission has not demonstrated that the conduct of those direct partners was different from that of the other direct partners is irrelevant in that regard.

548    In the second place, first, it should be noted, as Google does, that [confidential] had stated, in reply to a request for information of the Commission, that it ‘ha[d] selected to work with one provider for any given ad type because it d[id] not believe that adding an additional [intermediary would] improve [its] product or increase its monetization’.

549    It follows that [confidential] did not wish to source from different intermediaries simultaneously. That finding is not called into question by the fact, already cited by the Commission and recalled in paragraph 504 above, that [confidential] had accepted the exclusivity clause, on the ground that Google required it ‘as a condition’ for concluding the GSA.

550    The Commission does indeed argue that [confidential], which is a subsidiary of [confidential], had replied to it stating that it worked with a broad number of intermediaries.

551    However, on the one hand, it must be pointed out that [confidential] had specifically stated, in its reply, that the intermediaries it referred to were media agencies and not providers of online search advertising intermediation services. On the other hand, it must be pointed out that [confidential] had indicated to the Commission that it sourced exclusively from Google as regards text ads, which, as follows from recital 137 of the contested decision, ‘typically’ correspond to online search ads.

552    It follows that, contrary to what the Commission suggests, [confidential]’s reply is not such as to demonstrate that all sites direct partners wished to source from several intermediaries simultaneously.

553    Second, it should be noted, as Google does, that [confidential] had stated, in reply to a request for information of the Commission, that, ‘in general, selecting only one provider per type of device enable[d] getting better financial conditions’ and that, ‘in any event, from a business point of view, it d[id] not really make sense to have several providers of online search [advertising] intermediation at the same time on the same page/and type of devices’.

554    Contrary to what the Commission suggests, the fact that [confidential] had stated, in essence, that it was financially more advantageous to source, for each website, from a single intermediary tends to corroborate the fact that all sites direct partners, which had chosen to include typically all of their websites in their GSAs, would not have sourced part of their requirements from other intermediaries had it not been for the exclusivity clause.

555    Third, it should be noted that Surfboard claims that it had no economic incentive to display ads from different intermediaries on the same page and that it had historically preferred to work with only one intermediary at a time.

556    Contrary to what the Commission suggests, the fact that other all sites direct partners, the replies of which are quoted in recitals 367 and 368 of the contested decision, had been deterred by the exclusivity clause from sourcing from other intermediaries is not such as to demonstrate that Surfboard would have sourced from such intermediaries had it not been for the said clause. Likewise, the fact that Google had concluded GSAs containing the exclusivity clause with all sites direct partners and that that fact was an indication that Google considered that those direct partners had an economic incentive to source from other intermediaries does not allow Surfboard’s assertions to be called into question.

557    In that regard, it is appropriate to note that Google has succeeded in demonstrating that only [confidential], the [confidential] group and Surfboard preferred to source from one intermediary at a time. However, their replies cannot call into question the examples of replies of all sites direct partners set out in recitals 367 and 368 of the contested decision, which, as has been noted in paragraphs 518 and 519 above, were capable of corroborating the Commission’s assessment that the exclusivity clause in GSAs concluded with such direct partners could have deterred them from sourcing at least part of their requirements from Google’s competing intermediaries.

(5)    Conclusion on the deterrent effect of the exclusivity clause vis-à-vis all sites direct partners

558    It is apparent from the foregoing considerations that, subject to the examination of all the other relevant circumstances, particularly of the duration for which the exclusivity clause applied (see paragraph 562 below), the Commission was right to find that that clause, prohibiting all sites direct partners from displaying competing ads on the websites that were included in their GSAs, could have deterred certain of them from sourcing at least part of their requirements from Google’s competing intermediaries.

559    First, it should be noted that, as has been mentioned in paragraph 518 above, the Commission could rightly take into consideration the examples of replies of all sites direct partners set out in recitals 367 and 368 of the contested decision as elements capable of corroborating its assessment that the exclusivity clause in GSAs concluded with such direct partners could have deterred them from sourcing at least part of their requirements from Google’s competing intermediaries.

560    Second, it should be noted, as the Commission did in recitals 230, 276 and 364 of the contested decision, that Google’s market share had increased between 2006 and 2016 in the majority of the national online search advertising markets at issue and in the market for online search advertising intermediation. Thus, in 2016, almost none of Google’s competitors on those markets remained. In addition, those markets were characterised by the existence of significant barriers to entry and expansion and by a lack of countervailing buyer power on the part of advertisers and publishers. In particular, the effects of scale and network effects had rendered the emergence of new competitors difficult.

561    In those circumstances, it is appropriate to find that, contrary to what Google claims, the mere fact that the exclusivity clause in GSAs concluded with all sites direct partners had had an effect only on the conduct of certain of those direct partners does not suffice to demonstrate that that clause had not been capable of restricting competition.

562    Consequently, it is necessary to find that the exclusivity clause in GSAs concluded with all sites direct partners could have been capable of producing the foreclosure effect found in the contested decision. However, as has been recalled in paragraphs 389, 399 and 400 above, the question of whether that clause actually had such a capability depends also on the examination of all the other relevant circumstances and, in particular, of the duration for which those direct partners were obliged, in view of the said clause, to source all or most of their requirements exclusively from Google, as the Commission found, rightly, in recital 364 of the contested decision.

(b)    Impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

563    In the contested decision, the Commission considered that the exclusivity clause in GSAs concluded with all sites direct partners had prevented Google’s competitors from accessing a significant part of the market for online search advertising intermediation in the EEA. In that regard, first, it noted that the gross revenues generated by those GSAs represented a significant part of that market. Second, it found that Google systematically included the exclusivity clause in the GSAs which generated the most revenues for it. Third, it observed that the number of queries performed on the websites included in GSAs concluded with those direct partners was significant compared to the number of queries performed on competing general search services. Fourth, it took the view that the period during which the exclusivity clause required those direct partners to source all or most of their requirements from Google was long. Fifth, it noted that the fact that that clause prevented competing intermediaries from accessing a significant part of the said market was consistent with the evolution of Google’s market shares. Sixth, it found that the said clause covered some of the most visited websites in the EEA. Seventh, it considered that the exclusivity clause in GSAs concluded with 69 other direct partners, which had been unable to confirm to it whether or not they typically included all of their websites in those GSAs, had prevented competing intermediaries from providing their services on the websites included in those GSAs.

564    First, Google claims that the exclusivity clause did not apply to all online search ad formats. Second, it disputes the analysis of the coverage of the market by the exclusivity clause in GSAs concluded with all sites direct partners. Third, it takes issue with the Commission for not having shown that that clause was capable of excluding a competitor as efficient as it. Fourth, it argues that the Commission did not take into account the option, for all sites direct partners, to source from competing intermediaries at the end of either the initial term of each of their GSAs or of the possible extensions thereof, as well as where a unilateral termination right had been provided for.

565    It is appropriate to examine each of Google’s arguments separately.

(1)    Application of the exclusivity clause to certain online search ad formats

566    Google claims that the exclusivity clause did not prevent direct partners from displaying other online search ad formats, such as product listing ads (‘PLAs’), as well as online non-search ads.

567    The Commission disputes Google’s line of argument.

568    On the one hand, it is apparent from recital 28 of the contested decision, the content of which is not disputed by Google, that PLAs are the results of specialised searches. It must be noted, however, that, including in the context of the first plea, Google does not claim – let alone prove – that the results of specialised searches and online search ads were part of the same market. Thus, it is appropriate to find that Google does not prove that intermediation services linked to online search advertising and those linked to specialised search results belonged to the same market, either.

569    Moreover, Google does not explain which other online search ad formats, in addition to PLAs, direct partners could display and it has not proved that those formats belonged to the same market as that of online search ads.

570    On the other hand, it should also be held that, as follows from paragraph 305 above, Google has not proved that the two types of ad at issue, namely online search ads and online non-search ads, belonged to the same market, meaning that it has not proved that online search advertising intermediation services and online non-search advertising intermediation services belonged to the same market, either.

571    Accordingly, it is appropriate to hold that the fact that the exclusivity clause in GSAs concluded with all sites direct partners did not apply to certain online search ad formats – including PLAs – or to online non-search ads does not call into question the fact that that clause prevented Google’s competitors from accessing a significant part of the market for online search advertising intermediation.

(2)    Coverage of the market by the exclusivity clause in GSAs concluded with all sites direct partners

572    In the contested decision, first, at the outset, the Commission considered that, between 2006 and 2009, the gross revenues generated by GSAs concluded with all sites direct partners represented between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA. It specified in that regard that the GSAs concluded with [confidential] (which belonged to the [confidential] group and which together constitute one of the all sites direct partners identified by the Commission), [confidential], [confidential] and [confidential] alone represented between [confidential] and [confidential]% of that market. Next, it found that, between 2010 and 2012, the gross revenues generated by GSAs concluded with all sites direct partners represented between [confidential] and [confidential]% of the said market and that the gross revenues generated by GSAs containing the placement clause had, for their part, increased from [confidential] to [confidential]% in that same market. Last, it noted that, between 2013 and 2015, the gross revenues generated by GSAs concluded with all sites direct partners represented at least [confidential]% of the market in question, as it confirmed during the hearing, and that the gross revenues generated by GSAs containing the placement clause represented at least [confidential]% of that market.

573    Second, the Commission found that Google had systematically included the exclusivity clause in the GSAs generating the most revenues for it. It noted, in that regard, that, between 2006 and 2009 and between 2010 and 2012, all sites direct partners had represented between [confidential] and [confidential]% and between [confidential] and [confidential]% of the gross revenues generated by all direct partners, respectively.

574    Third, the Commission found that, while, in the EEA, the number of online queries carried out on the websites of all sites direct partners was significant compared to the number of online queries carried out on the general search services of Google’s competitors, it was still negligible compared to the number of online queries carried out on Google’s general search service.

575    Fourth, the Commission noted that it followed from the evolution of Google’s market share in the market for online search advertising intermediation in the EEA that the exclusivity clause in GSAs concluded with all sites direct partners had prevented competing intermediaries from accessing a significant part of that market.

576    Fifth, the Commission noted that, according to a study submitted in 2011 and updated in 2013 by Microsoft (‘the Microsoft study’), which is one of the complainants in the present case, the exclusivity clause in GSAs concluded with all sites direct partners covered some of the most visited websites. More specifically, it found that Google had provided online search advertising intermediation services to between [confidential] and [confidential]% of the most visited web domains in France, Germany, Italy, Spain and the United Kingdom in 2010.

577    Sixth, the Commission recalled that, between 2006 and 2009, the exclusivity clause was also present in GSAs concluded with 69 other direct partners which had been unable to confirm whether they had typically included all of their websites in their GSAs. It infers from this that that clause had prevented Google’s competitors from providing online search advertising intermediation services for the websites included in those GSAs during that period.

578    Google maintains that the analysis of the market coverage by the exclusivity clause in GSAs concluded with all sites direct partners, based on the proportion of revenues generated by those GSAs, is irrelevant for the purpose of showing that the exclusivity clause had a foreclosure effect. Furthermore, it criticises the Commission (i) for having found that certain of those direct partners sourced all or most of their requirements exclusively from it, (ii) for having taken into account the revenues of direct partners which were not all sites direct partners, (iii) for having taken an approach in relation to the exclusivity clause that was inconsistent with that taken in relation to the placement and prior authorisation clauses, (iv) for having combined the coverage of the exclusivity clause with that of the placement and prior authorisation clauses, and (v) for not having determined the coverage of GSAs concluded with all sites direct partners for 2016.

(i)    Taking into account of data subsequent to the conclusion of GSAs for the purpose of calculating the coverage rate of the exclusivity clause

579    Google claims that direct partners were free to choose the websites that they wished to include in their GSAs. An ex post analysis of the coverage rate of those GSAs thus shows the outcome of competition on the merits, but does not evidence the foreclosure effect of the exclusivity clause.

580    The Commission disputes Google’s line of argument.

581    In that regard, it is appropriate to recall, as the Commission does, that exclusive supply obligations are designed to deprive the purchaser of or restrict his, her or its possible choices of sources of supply and to deny other producers access to the market (see, to that effect, judgment of 13 February 1979, Hoffmann-La Roche v Commission, 85/76, EU:C:1979:36, paragraph 90).

582    In the case at hand, it should be noted, as has been mentioned in paragraph 414 above, that once a direct partner had chosen to include one of its websites in its GSA, it necessarily had to source its requirements in terms of online search advertising intermediation services exclusively from Google as far as that website was concerned. As a result, where a publisher had chosen to include all of its websites in its GSA, pursuant to the exclusivity clause, it necessarily had to source all of its requirements in terms of online search advertising intermediation services exclusively from Google for the duration of that GSA. Thus, Google’s competitors were denied the possibility of providing their services to that publisher for that duration.

583    In those conditions, it is appropriate to find that, contrary to what Google suggests, the fact that the revenues generated by GSAs in which direct partners had typically included all of their websites represented a significant part of the online search advertising intermediation market was capable of producing a foreclosure effect contrary to Article 102 TFEU vis-à-vis Google’s other competing intermediaries, even if those direct partners themselves had initially chosen the websites that they had included in their GSAs.

(ii) Taking into account of the revenues generated by GSAs concluded with direct partners belonging to the same group as certain all sites direct partners

584    Google claims that, even assuming that the all sites direct partners identified in the contested decision had sourced all or most of their requirements exclusively from Google, the Commission was wrong to take account of not only the revenues generated by GSAs concluded by those direct partners, but also those generated by contracts not containing the exclusivity clause concluded by other entities of the groups to which the latter belonged. It follows that the Commission artificially inflated the coverage of the exclusivity clause in GSAs concluded with all sites direct partners.

585    By way of example, Google notes that, according to the contested decision, [confidential] and [confidential], all part of the [confidential] group, each constituted separate direct partners included among the all sites direct partners. However, it asserts that other entities of that group, including [confidential] and [confidential], had concluded contracts with it relating to the provision of AFS and that the Commission had taken account of the revenues generated by those contracts for the purposes of calculating the revenues generated by GSAs concluded by all sites direct partners.

586    The Commission disputes Google’s line of argument.

587    In that regard, it should be noted, as the Commission does, that Google produces no evidence in support of its claim that the contested decision took into account the revenues generated by the contracts concluded with [confidential] and [confidential] in order to calculate the revenues generated by the [confidential] group.

588    In particular, it should be noted that, in response to a measure of organisation of procedure requesting Google to provide the contracts that it had concluded with [confidential] and [confidential] relating to the use of AFS, Google merely stated that [confidential] and [confidential] had not concluded GSAs with it. It nevertheless failed to provide evidence that [confidential] and [confidential] had in fact used AFS by means, for example, of online contracts. Thus, it did not prove that those contracts had generated revenues which the Commission thereafter took into account, wrongly, to determine the coverage rate of the exclusivity clause in GSAs concluded with all sites direct partners.

589    In those conditions, and in so far as Google does not cite other examples capable of proving that the Commission used data concerning the revenues generated by contracts concluded with publishers which were not all sites direct partners, it is appropriate to find that Google is not justified in criticising the Commission for having taken into account the revenues generated by such contracts.

(iii) Taking into account of GSAs containing placement and prior authorisation clauses

590    Google criticises the Commission for having taken into account, for the purposes of assessing the effects of the exclusivity clause in GSAs concluded with all sites direct partners, the revenues generated by GSAs containing the placement and prior authorisation clauses, including, moreover, where those direct partners that had concluded those GSAs had not included all of their websites in them.

–       Taking into account of the revenues generated by GSAs containing the placement and prior authorisation clauses for the purposes of assessing the foreclosure effects of the exclusivity clause in GSAs concluded with all sites direct partners

591    As follows from paragraph 572 above, the Commission took into account, in the contested decision, the gross revenues generated by GSAs containing the placement clause to find that the gross revenues generated by GSAs concluded with all sites direct partners represented a significant part of the market for online search advertising intermediation.

592    Google criticises the Commission for having taken into account, to characterise the foreclosure effect of the exclusivity clause in GSAs, the revenues generated by GSAs containing placement and prior authorisation clauses. In that regard, it states that, in so far as the Commission identified, in the contested decision, three separate infringements of Article 102 TFEU, resulting from the inclusion of the three clauses – exclusivity, placement and prior authorisation, respectively – the Commission had to take into account the specific coverage of each of those clauses.

593    The Commission disputes Google’s line of argument.

594    First, it should be pointed out that, contrary to what Google suggests, the Commission did not specifically take into account the revenues generated by GSAs containing the prior authorisation clause, even if it is not disputed that, as is indicated in recital 630, point 4, of the contested decision, all GSAs containing the prior authorisation clause also contained the placement clause.

595    Second, it is necessary to hold that, as follows from paragraphs 107 and 390 above, the Commission had to take into account, in its assessment of the effects of the exclusivity clause, all the circumstances of the case and, in particular, the rate of market coverage of the exclusivity clause (see, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 139).

596    In addition, the foreclosure by a dominant undertaking of a substantial part of the market cannot be justified by showing that the contestable part of the market is still sufficient to accommodate a limited number of competitors. First, the customers on the foreclosed part of the market should have the opportunity to benefit from whatever degree of competition is possible on the market and competitors should be able to compete on the merits for the entire market and not just for a part of it. Second, it is not the role of the dominant undertaking to dictate how many viable competitors will be allowed to compete for the remaining contestable portion of demand (judgment of 19 April 2012, Tomra Systems and Others v Commission, C‑549/10 P, EU:C:2012:221, paragraph 42).

597    In that context, in order to determine whether the coverage rate of the exclusivity clause was sufficient, in the circumstances of the case, to enable that clause to produce a foreclosure effect, the Commission necessarily had to determine the share of the market for online search advertising intermediation to which Google’s competitors could have had access had it not been for the said clause.

598    On the one hand, however, as has essentially been noted in paragraphs 78 to 82 above, the Commission considered that the placement clause was capable of creating a foreclosure effect, by limiting, at least to a certain extent, the ability of direct partners to display competing online search ads.

599    More specifically, the Commission found, in recital 630 of the contested decision, that the exclusivity and placement clauses were complementary in that they sought to deter direct partners from sourcing competing search ads and to prevent access by Google’s competing intermediaries to a significant part of the market for online search advertising intermediation in the EEA. It also recalled, in recitals 335, 467, 630, point 2, 712 and 718 of the contested decision, that Google itself had described the placement clause as a relaxed exclusivity clause. It thus considered that clause capable of restricting the part of the market on which Google and its competitors could compete.

600    It follows that the placement clause was, according to the Commission, capable of restricting the part of the market for online search advertising intermediation to which Google’s competitors were able to have access.

601    On the other hand, it is apparent from recitals 89 and 335 of the contested decision that Google had progressively started to replace, from March 2009, the exclusivity clause particularly with the placement clause in its GSAs, such that those clauses could cover different parts of the market for online search advertising intermediation simultaneously, overlapping, at least in part, for the period between March 2009 and 31 March 2016.

602    Google’s proposition, the essence of which is that the Commission ought to have examined the coverage of the exclusivity clause in GSAs concluded with all sites direct partners and that of the placement clause in isolation from each other, would however be tantamount to partitioning artificially the examination of the coverage of the market for online search advertising intermediation according to the practices alleged against Google, with no regard for their factual and legal context, which is characterised in particular by the gradual replacement of the exclusivity clause with the placement clause. Such a partition would have the logical consequence of lowering the coverage rate of the exclusivity clause in GSAs concluded with all sites direct partners during the latter years in which that clause was applied, all the while ignoring the fact that, during those same years, the placement clause had quickly exceeded the part of the market covered by the said exclusivity clause as from the amendment of the template GSA in March 2009. Such a partition would thus not reflect the economic reality of that market between 2009 and 2016.

603    In those conditions, and subject to the question, raised in the third plea, of whether the placement clause was in fact capable of producing a foreclosure effect contrary to Article 102 TFEU, the Commission cannot be considered to have committed an error of law  on the sole ground that it took into account the coverage of the placement clause in order to determine whether the coverage rate of the exclusivity clause had been sufficient to prevent Google’s competitors from accessing a significant part of the market for online search advertising intermediation.

–       Taking into account of the revenues generated by GSAs containing the placement and prior authorisation clauses concluded with direct partners that typically did not include all of their websites in those GSAs for the purposes of assessing the foreclosure effects of the exclusivity clause in GSAs concluded with all sites direct partners

604    In recitals 523 and 593 of the contested decision, the Commission explained that it had reduced to 34 the number of direct partners that it considered to source all or most of their requirements exclusively from Google, whereas 69 other direct partners (mentioned in paragraph 563 above), had included at least some of their websites displaying online search ads in their GSAs containing the exclusivity clause. It inferred from this that Google was not justified in criticising it for having taken into account every GSA containing the placement and prior authorisation clauses, and not only the GSAs in which direct partners had typically included all of their websites.

605    Google claims that the Commission’s analysis was inconsistent. It asserts that that institution considered the exclusivity clause contrary to Article 102 TFEU only to the extent that that clause was contained in GSAs concluded with all sites direct partners, whereas the placement and prior authorisation clauses were considered contrary to that provision since they were contained in any GSA. It criticises, moreover, the Commission for having taken into account, for the purposes of the assessment of the coverage of the exclusivity clause in GSAs concluded with all sites direct partners, the revenues generated by all GSAs containing the placement and prior authorisation clauses, including those in which direct partners had not typically included all of their websites.

606    The Commission disputes Google’s line of argument.

607    First, it should be noted that it is apparent from recital 349 of the contested decision that the 69 direct partners to which the Commission refers in recitals 523 and 593 of the contested decision are those which were not in a position to confirm whether or not they had typically included all of their websites displaying online search ads in their GSAs. Those are thus direct partners for which the Commission had not managed to prove, pursuant to the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36), that they sourced all or most of their requirements exclusively from Google.

608    Consequently, although, in recital 349 of the contested decision, the Commission explained that it had taken a conservative approach that was favourable to Google, it is only because it was possible that a larger number of direct partners were capable of sourcing all or most of their needs exclusively from Google. However, it cannot be inferred from that recital that the Commission could have found that the exclusivity clause in GSAs in which direct partners had typically not included all of their websites was contrary to Article 102 TFEU pursuant to the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36).

609    It follows that, irrespective of whether the Commission’s approach was ‘unduly conservative and favourable to Google’, as that institution indicates in paragraph 49 of its defence, the fact remains that the Commission did not find that the exclusivity clause in the GSAs concluded with the 69 direct partners referred to in paragraph 607 above was contrary to Article 102 TFEU.

610    Second, it should be noted, as the Commission did in recital 455 of the contested decision, that the placement and prior authorisation clauses were less restrictive than the exclusivity clause for direct partners in so far as they authorised them, at least to a certain degree, to use AFS and a competing online search advertising intermediation service simultaneously on websites included in their GSAs. Moreover, in recitals 335, 467, 630, point 2, 712 and 718 of the said decision, the Commission recalled a statement of Google describing the placement clause as a relaxed exclusivity clause.

611    In those conditions, it is appropriate to note, as Google does, that there is indeed a certain asymmetry to the contested decision, in so far as the Commission found the exclusivity clause contrary to Article 102 TFEU only to the extent that it was contained in GSAs concluded with all sites direct partners, when it found the placement and prior authorisation clauses contrary to that same provision since they applied to all direct partners whose GSAs contained such clauses.

612    However, it should be noted that, in order to determine whether the exclusivity clause was contrary to Article 102 TFEU, the Commission was not required to prove that that clause actually obliged all sites direct partners to source all or most of their requirements exclusively from Google within the meaning of the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36).

613    Indeed, in accordance with the case-law recalled in paragraph 108 above, the Commission could, inter alia, simply show that the exclusivity clause was capable of producing a foreclosure effect.

614    To that end, the Commission had to take into account all the circumstances of the case, in accordance with the case-law recalled in paragraph 595 above. In order to determine whether the coverage rate of the exclusivity clause in GSAs concluded with all sites direct partners was sufficient for it to have the capability to restrict competition, the Commission could take into account, as a relevant circumstance, the fact that the part of the market for online search advertising intermediation that was not covered by the said clause was partly covered by the placement clause, which limited the possibilities for Google’s competing intermediaries to access that market.

615    In those conditions, it is appropriate to find that the circumstance that the Commission only found the exclusivity clause contrary to Article 102 TFEU in so far as that clause was contained in GSAs concluded with all sites direct partners was not such as to prevent the Commission from finding the placement and prior authorisation clauses also contrary to that provision where they applied to all of the direct partners whose GSAs contained such clauses.

616    On the one hand, it follows that the contested decision cannot be regarded as being contradictory on the sole ground that the Commission found the exclusivity clause contrary to Article 102 TFEU only to the extent that that clause was contained in GSAs concluded with all sites direct partners, when it found the placement and prior authorisation clauses contradictory to that provision where they applied to all of the direct partners whose GSAs contained such clauses.

617    On the other hand, subject to the question, raised in the context of the third plea, of whether the placement clause was in fact capable of producing a foreclosure effect contrary to Article 102 TFEU, it is necessary to find that Google is not justified in arguing that the Commission took into account, wrongly, GSAs containing the placement clause, including those in which direct partners had not included all of their websites, in order to assess the extent of the coverage rate of the exclusivity clause.

(iv) Taking into account of direct partners that had not included all of their websites in their GSAs containing the exclusivity clause for the purpose of calculating the coverage rate of that clause

618    Google disputes the ‘all sites direct partners’ categorisation attributed by the Commission to certain operators. It claims that the Commission did not establish that all sites direct partners, identified in the contested decision, had included all of their websites in their GSAs and that they had consequently sourced all or most of their requirements exclusively from it. It infers from this that the Commission overstated the revenues generated by GSAs concluded with direct partners sourcing all or most of their requirements exclusively from it. Moreover, it argues that, even relying on the data contained in the contested decision, the Commission should have taken into account the fact that a significant part of the revenues of online search advertising intermediation services in the EEA remained ‘open’ to the other intermediaries.

619    It is appropriate to note that, in recital 395 of the contested decision, the Commission considered irrelevant the question whether ‘considerable’ revenues generated in the EEA by online search advertising intermediation services remained ‘available’ to Google’s competitors. It considered it sufficient that the gross revenues generated by GSAs concluded with all sites direct partners had represented a significant part of the market for those services between 2006 and 2015.

620    First, it is appropriate to note that, contrary to what Google suggests, the part of the online search advertising intermediation market that was not covered by the exclusivity clause in GSAs concluded with all sites direct partners did not necessarily remain ‘open’ to Google’s competitors. It should be noted that, as the Commission held in the contested decision, from March 2009, that part of the market was partly covered by the placement clause, which progressively replaced the exclusivity clause from that date. As follows from paragraph 598 above and from paragraph 767 below, the placement clause reserved the most prominent spaces of results pages for Google’s ads.

621    Second, it must be stated that, as has been recalled in paragraph 596 above, the foreclosure by a dominant undertaking of a substantial part of the market cannot be justified by showing that the contestable part of the market is still sufficient to accommodate a limited number of competitors.

622    Consequently, as the Commission contends, the circumstance that a significant part of the online search advertising intermediation market was covered by the exclusivity clause in GSAs concluded with all sites direct partners could, in principle, be sufficient to produce a foreclosure effect.

623    In that regard, the Commission found, in Table 13 of the contested decision, that the coverage rate of the exclusivity clause in GSAs concluded with all of the all sites direct partners identified in that decision was the following:

624    As has been noted in paragraph 450 above, Google and Surfboard claim that 11 of the 34 all sites direct partners identified in the contested decision did not source at a minimum most of their requirements exclusively from Google. Those are [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], Surfboard, [confidential], [confidential], [confidential] and [confidential].

625    In the interest of procedural economy, the Court finds that, under the scenario most favourable to Google, namely that which assumes that the coverage rate corresponding to the GSAs concluded with the 11 direct partners identified by Google and Surfboard should be deducted from the coverage rate found in the contested decision, it would follow from Table 13 of the contested decision that the coverage rate of the exclusivity clause in GSAs concluded with the other all sites direct partners identified in the contested decision would be the following:

626    In that regard, it is appropriate to hold that, under that scenario most favourable to Google, the relatively low coverage rate of the exclusivity clause in GSAs concluded with all sites direct partners, particularly as from 2010, results, having regard to the considerations set out in paragraph 601 above, inter alia from the progressive replacement of the exclusivity clause with the placement clause in GSAs following the amendment of the template GSA in March 2009.

627    In addition, it is settled that the gross revenues generated by GSAs containing the placement clause, which was introduced to it only as from March 2009, represented between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA between 2009 and 2015.

628    In the contested decision, the Commission found, in that regard, in Table 25 of that decision, that the combined coverage rate of the exclusivity clause in GSAs concluded with all of the all sites direct partners identified in that decision, on the one hand, and of the placement clause, on the other hand, was the following:

629    Nevertheless, based on the scenario most favourable to Google, namely that in which the coverage rate of the exclusivity clause in GSAs concluded with the all sites direct partners identified would have been, between 2006 and 2012, that indicated in paragraph 625 above and would have been zero between 2013 and 2015, which is however not alleged by Google, it would follow from Table 24 of the contested decision that the coverage rate of the placement clause, combined with that of the exclusivity clause, would have been the following:

630    It follows that, even on the basis of the data most favourable to Google, on the one hand, in the period preceding the introduction of the placement clause to GSAs, between 2006 and 2008, the coverage rate of the exclusivity clause in GSAs in which direct partners had typically included all of their websites was between [confidential]%, in 2007, and [confidential]%, in 2008. On the other hand, in the period during which the placement clause applied, that is to say, between 2009 and 2015, the combined coverage rate of those two clauses was between [confidential]%, in 2009, and [confidential]%, in 2012.

631    In those conditions, it is appropriate to find that, even in the scenario most favourable to Google, the combined coverage rate of the exclusivity clause in GSAs in which direct partners had typically included all of their websites, on the one hand, and the placement clause, on the other hand, could be sufficient for those clauses to be capable of producing a foreclosure effect between 2006 and 2015.

632    It follows that Google is not justified in arguing that the exclusivity clause in GSAs in which direct partners had typically included all of their websites was not capable of producing a foreclosure effect on the sole ground that a significant part of the market for online search advertising intermediation was not covered by those clauses between 2006 and 2015.

(v)    Revenues generated by GSAs concluded with all sites direct partners in 2016

633    Google criticises the Commission for not having evaluated the coverage of the exclusivity clause in GSAs concluded with all sites direct partners and that of the placement and prior authorisation clauses in relation to 2016.

634    First, the Commission contends that it follows from recitals 388 and 457 and from Table 15 of the contested decision that [confidential], 1 of the 3 most important direct partners of the 34 all sites direct partners identified in the contested decision, remained party to a GSA containing an exclusivity clause until 31 March 2016. Second, it notes that it follows from recitals 99 to 106, 564 and 633 of that decision that, on 6 September 2016, Google informed the last direct partner of its decision to waive the placement clause. It adds that a certain number of direct partners, including significant direct partners, were party to a GSA containing that clause until 3 June 2016.

635    In that regard, it should be recalled, as has been noted in paragraph 594 above, that the Commission did not specifically take into account the revenues generated by GSAs containing the prior authorisation clause in the assessment of the coverage of the exclusivity clause in GSAs concluded with all sites direct partners, even if it is not disputed that, as Google explains, all GSAs containing the prior authorisation clause also contained the placement clause.

636    However, it must be noted, as Google does, that the contested decision does not identify the part of the market for online search advertising intermediation covered by the exclusivity clause in GSAs concluded with all sites direct partners, on the one hand, and that of the placement clause, on the other hand, for 2016, whereas it considered that each of those clauses constituted an abuse of a dominant position until 31 March 2016 and 6 September 2016, respectively.

–       Exclusivity clause in GSAs concluded with all sites direct partners

637    In respect of the exclusivity clause in GSAs concluded with all sites direct partners, it must be pointed out that the Commission merely refers to the recitals of the contested decision from which it is apparent that the last GSA containing that clause in which one of those direct partners – [confidential] – had typically included all of its websites had expired on 31 March 2016.

638    However, while the Commission asserts, in the defence, that [confidential] was ‘one of the three most important [direct partners]’ of the all sites direct partners, suffice it to note that it did not present, in the contested decision, evidence allowing for assessment of the amount of the revenues generated by the GSA at issue concluded with that direct partner in relation to 2016 specifically. In particular, it is limited to stating, in the said decision, that [confidential]’s websites represented, on average, [confidential]% of the gross revenues generated by Google in the EEA market for online search advertising intermediation services between 2006 and 2012, even though it also noted that the exclusivity clause had applied to those websites between 15 May 2003 and 31 March 2016.

639    In those conditions, given that the Commission identified no other direct partner, other than [confidential], that had typically included all of its websites in a GSA containing the exclusivity clause in the course of the period between 1 January and 31 March 2016, it must be held that it likewise did not prove that, owing to its coverage, that clause could produce a foreclosure effect during that period, irrespective of whether, as Google claims, the said clause, as it was drafted in the GSAs mentioned in recital 348 of the contested decision, had obliged [confidential] to source all or most of its requirements exclusively from Google.

–       Placement clause

640    As far as the placement clause is concerned, it should be noted that the Commission is limited to stating, in the contested decision, that Google had informed it, first, on 28 May 2016, that it was intending to amend the said clause and, second, on 9 September 2016, that it had sent [confidential] letters to all direct partners to that end. However, the Commission did not present, in that decision, evidence allowing the extent of the coverage of GSAs containing the clause in question in relation to 2016 to be assessed. In that regard, it must be held that the Commission’s assertion formulated in the defence, according to which nine direct partners were still subject to such GSAs until 3 June 2016, is not such as to call that finding into question.

641    In those conditions, it is appropriate to find that the Commission did not establish that the exclusivity and placement clauses could have prevented Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA in 2016.

(vi) Traffic of the websites covered by the exclusivity clause in GSAs concluded with all sites direct partners

642    Google claims that the Microsoft study, mentioned in paragraph 576 above, cannot be used to assess the effects of the exclusivity clause in GSAs concluded with all sites direct partners in so far as, on the one hand, certain of the websites identified in that study were not subject to that clause and, on the other hand, the period assessed and the number of Member States were limited. In addition, it argues that the number of website visits is not necessarily a reliable indication of the number of online queries performed on that website or, consequently, of the revenues from online search advertising generated by that website. It notes, in that regard, that it is apparent from an update subsequent to the said study that websites using Microsoft’s services received more visits than those using Google’s services.

643    The Commission disputes Google’s line of argument.

644    In recital 390 of the contested decision, the Commission found, on the basis of the Microsoft study, that ‘some’ of the most visited websites in the EEA were covered by the exclusivity clause in GSAs concluded with all sites direct partners.

645    In that regard, it should be noted that, as follows from paragraph 623 above, the Commission determined, in the contested decision, the exact coverage rate of the exclusivity clause in GSAs in which direct partners had typically included all of their websites on the basis of the gross revenues generated by those GSAs.

646    The Microsoft study, mentioned in recital 390 of the contested decision, establishes that Google provided online search advertising intermediation services to between [confidential] and [confidential]% of the most visited web domains in Germany, Spain, France, Italy and the United Kingdom in 2010. It is true that, as Google notes, that study concerns only one year of the period of infringement and five Member States. Likewise, Google is justified in noting that that study does not enable the traffic generated specifically by the websites covered by the exclusivity clause in GSAs concluded with all sites direct partners to be identified. Last, it rightly observes that the number of website visits is not necessarily a reliable indication of online search advertising revenues.

647    However, the fact remains that the Microsoft study constitutes an additional indication enabling an assessment of the scale of the online search advertising intermediation services provided by Google as well as the coverage of the market for online search advertising intermediation by that clause, in so far as, first of all, it involves five of the largest Member States of the EEA, next, it is not disputed that at least some of the websites covered by that study were subject to the exclusivity clause in GSAs concluded with all sites direct partners and, last, there is some correlation between the number of website visits and online search advertising revenues.

648    In those conditions, it is appropriate to find that the Commission could rely on the Microsoft study as an indication corroborating the examination of the coverage rate, without it being necessary to rule on the admissibility of Google’s line of argument, disputed by the Commission.

649    In any event, the Commission did not base its calculation of the coverage rate of the exclusivity clause in GSAs concluded with all sites direct partners on the Microsoft study, such that, even assuming that Google’s arguments in that regard were founded, they would have no effect on the calculation of that rate performed by the Commission.

(vii) Conclusion on the coverage of the market by the exclusivity clause in GSAs concluded by all sites direct partners

650    In view of the foregoing, it is appropriate to find that, even in the scenario most favourable to Google, the Commission could rightly consider that, having regard to the coverage of the placement clause in the light of the circumstances recalled in paragraph 602 above, the coverage of the exclusivity clause in GSAs in which direct partners had typically included all of their websites could be sufficient to enable that clause to have the capacity to produce a foreclosure effect between 1 January 2006 and 31 December 2015. However, it must be held that the Commission did not establish that the exclusivity clause could have produced such an effect, owing to its coverage, between 1 January and 31 March 2016.

(3)    As-efficient competitor test

651    In recital 433 of the contested decision, the Commission noted that the exclusivity clause in GSAs concluded with all sites direct partners was capable of excluding a hypothetical competitor as efficient as Google. First, the revenues generated by those GSAs between 2006 and 2009 represented between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA. Second, the revenues generated by the said GSAs and those containing the placement clause between 2009 and 2015 represented between [confidential] and [confidential]% of that market. Third, Google held a ‘very large’ share of the market between 2006 and 2016. Fourth, that market was prone to network effects.

652    The Commission also indicated, in recital 434 of the contested decision, that it was ‘doubtful’ whether a hypothetical intermediary as efficient as Google could have emerged during the period of application of the exclusivity clause in GSAs concluded with all sites direct partners. It last found that the question of whether Google had pursued a strategy aiming at excluding competitors as competitive as it was irrelevant.

653    Google argues that the Commission failed to prove that a competitor as efficient as it could not have emerged on the market for online search advertising intermediation, or that such a competitor was capable of being excluded from that market. Surfboard adds that that clause did not prevent it from sourcing competing ads where it considered that the services of a competing intermediary were attractive. It notes that it in fact sourced from Yahoo! for one of its websites.

(i)    Preliminary observations

654    Google claims that the Commission was limited to considering, wrongly, that it was ‘doubtful’ that a hypothetical intermediary as efficient as it could have emerged in the course of the period during which the exclusivity clause in GSAs concluded with all sites direct partners was applicable. It is of the view that the Commission should have proved that restriction of competition was ‘likely’.

655    The Commission disputes Google’s line of argument.

656    In that regard, it should be noted that the Commission, in recital 433 of the contested decision, concluded, on the basis of the four factors enumerated in paragraph 651 above, that the exclusivity clause had been ‘capable’ of foreclosing a hypothetical competitor as efficient as Google and that, in recital 434 of that decision, it notes, on the basis of those same factors, that it was ‘doubtful’ that such a competitor could have emerged during the period of application of the exclusivity clause contained in GSAs concluded with all sites direct partners.

657    In those conditions, Google’s line of argument should be regarded as being purely terminological in nature and the Commission cannot be criticised for having found that conduct which was ‘capable’ of foreclosing a hypothetical competitor as efficient as Google rendered the emergence of such a competitor ‘doubtful’.

(ii) Factors relevant to the application of the as-effective competitor test

658    Google claims that none of the factors mentioned in recital 433 of the contested decision demonstrate that it was ‘practically impossible’ for a competitor as efficient as it to emerge in the present case or that one was likely to be foreclosed. Thus, first, it is of the view that it was impossible to determine from the coverage of the exclusivity clause in GSAs concluded with all sites direct partners in itself whether such a competitor was able to compete ‘profitably’ and that, in any event, that competitor could, at all times, access a significant part of the market at issue. Second, it argues that the Commission artificially inflated the scope of that coverage by taking into account the revenues from direct partners subject to the placement clause that had not included all of their websites in their GSAs. Third, it criticises the Commission, on the one hand, for having overstated its market shares and, on the other hand, for not having established a causal link between its market shares, the ‘position of AECs [as-efficient competitors]’ and the said exclusivity clause. Fourth, it considers that the Commission did not establish that the circumstance that the market at issue was prone to network effects showed that a competitor as efficient as it could not penetrate it.

659    The Commission disputes Google’s line of argument.

660    In that connection, it is appropriate to recall that the as-efficient competitor test consists in examining whether the practices of a dominant undertaking could drive an equally efficient competitor from the market (see, to that effect, judgment of 6 October 2015, Post Danmark, C‑23/14, EU:C:2015:651, paragraph 53). That concept thus refers, in practice, to various tests which have in common the aim of assessing the ability of a practice to produce anticompetitive exclusionary effects by reference to the ability of a hypothetical competitor of the undertaking in a dominant position, which is as efficient as the dominant undertaking in terms of cost structure, to offer customers a rate which is sufficiently advantageous to encourage them to switch supplier, despite the disadvantages caused, without that causing that competitor to incur losses. That ability is generally determined in the light of the cost structure of the undertaking in a dominant position itself (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 56).

661    A test of that nature may be inappropriate in particular in the case of certain non-pricing practices, such as a refusal to supply, or where the relevant market is protected by significant barriers. Such a test is only one of a number of methods for assessing whether a practice is capable of producing exclusionary effects; moreover, that method takes into consideration only price competition. In particular, the use by an undertaking in a dominant position of resources other than those governing competition on the merits may be sufficient, in certain circumstances, to establish the existence of such an abuse (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 57).

662    Nevertheless, even in the case of non-pricing practices, the relevance of such a test cannot be ruled out. A test of that type may prove useful since the consequences of the practice in question can be quantified. In particular, in the case of exclusivity clauses, such a test may theoretically serve to determine whether a hypothetical competitor with a cost structure similar to that of the undertaking in a dominant position would be able to offer its products or services otherwise than at a loss or with an insufficient margin if it had to bear the compensation which the distributors would have to pay in order to switch supplier, or the losses which they would suffer after such a change following the withdrawal of previously agreed discounts (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 59).

663    Consequently, where an undertaking in a dominant position suspected of abuse provides the Commission with an analysis based on the as-efficient competitor test, that institution cannot disregard that evidence without even examining its probative value (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 60).

664    Be that as it may, it is not possible to infer from Article 102 TFEU or the case-law of the Court of Justice that there is a legal obligation requiring a finding to the effect that a practice carried out by a dominant undertaking is abusive to be based always on the as-efficient competitor test (see, to that effect, judgment of 6 October 2015, Post Danmark, C‑23/14, EU:C:2015:651, paragraph 57).

665    It follows that the Commission could merely demonstrate the capability of the exclusivity clause in GSAs concluded with all sites direct partners to produce a foreclosure effect by relying on several relevant elements, without necessarily relying, to that end, on the as-efficient competitor test.

666    In the case at hand, the very substance of the exclusivity clause, which prevented, in principle, all sites direct partners from displaying competing ads, and the factors mentioned in recital 433 of the contested decision, namely, on the one hand, the fact that that clause, together with the placement clause, covered a significant part of the online search advertising intermediation market, as has been noted in paragraph 650 above, and, on the other hand, the extent of Google’s dominant position resulting in particular from its very high market shares and from the barriers to entry and expansion in the form inter alia of network effects, could demonstrate that the exclusivity clause in GSAs concluded with all sites direct partners could be capable of foreclosing a hypothetical competitor as efficient as Google. It follows that those factors could also demonstrate that it was ‘doubtful’ that such a competitor could have emerged during the period of application of the said clause.

667    In those circumstances, and in accordance with the case-law recalled in paragraph 663 above, if Google had produced, during the administrative procedure, an analysis based on an as-efficient competitor test, within the meaning of the case-law recalled in paragraph 660 above, it was for the Commission to examine that analysis, which it is appropriate to verify below.

(iii) Evidence submitted by Google during the administrative procedure

668    Google claims that the Commission did not take into account the evidence that it submitted during the administrative procedure aimed at demonstrating that a competitor as efficient as it could emerge. Specifically, it notes that direct partners chose to procure online search ads from some of its competitors.

669    The Commission disputes Google’s line of argument.

670    In that regard, it must be pointed out that Google had identified, in response to the statement of objections, nine bids, organised by various publishers, including [confidential] and [confidential], that it had lost to [confidential], [confidential] or Microsoft, between the last quarter of 2006 and the second quarter of 2007 as well as in 2015 and 2016. Moreover, it appears to follow from Annex A.45 to the application that the Commission’s file contained evidence showing that [confidential] had continued to source part of its requirements from other intermediaries between 2006 and 2015 having managed to reach, in 2008, up to [confidential]% of its requirements.

671    Thus, while it is indeed true that, during the administrative procedure, Google had provided the Commission with a few one-off and isolated examples where direct partners had preferred to source from other intermediaries rather than from it, the fact remains that such examples do not constitute an analysis based on the as-efficient competitor test, within the meaning of the case-law recalled in paragraph 663 above. Moreover, it is settled that Google did not provide during the administrative procedure – nor has it provided before the Court – any analysis within the meaning of that case-law.

672    In any event, it must be held that, without additional details, the one-off examples given by Google do not suffice to prove that competitors at least as efficient as it could have emerged.

673    Furthermore, it should be recalled, as has been noted in paragraph 108 above, that, in order to establish that conduct is abusive, the Commission does not necessarily have to demonstrate that that conduct actually produced anticompetitive effects. The purpose of Article 102 TFEU is to penalise abuse by one or more undertakings of a dominant position within the internal market or in a substantial part of it, irrespective of whether such practice has proved successful (judgment of 19 January 2023, Unilever Italia Mkt. Operations, C‑680/20, EU:C:2023:33, paragraph 41).

674    Consequently, the fact that certain of Google’s competing intermediaries were able to win a few bids organised by publishers is not such as to rule out that the exclusivity clause in GSAs concluded with the 34 all sites direct partners was capable of producing a foreclosure effect contrary to Article 102 TFEU. Last, the fact that direct partners chose not to use AFS, to use it without concluding a GSA or to include only the websites they wished in their GSAs cannot call into question the Commission’s assessment that the exclusivity clause in GSAs concluded with all sites direct partners was capable of producing a foreclosure effect vis-à-vis a competitor at least as efficient as Google. Indeed, as follows from paragraph 582 above, once those GSAs were concluded, the said clause was capable of preventing Google’s competing intermediaries from providing their services to the said direct partners for the duration of those same GSAs.

675    In those conditions, the fact that Google lost certain bids in the course of the period of infringement cannot call into question the Commission’s assessment that the exclusivity clause in GSAs concluded with all sites direct partners was capable of producing a foreclosure effect vis-à-vis a competitor at least as efficient as Google.

(iv) Existence of a strategy aimed at excluding as-efficient competitors

676    Google criticises the Commission for having considered that the absence of a strategy aimed at excluding competitors as efficient as it was irrelevant in this case.

677    The Commission disputes Google’s line of argument.

678    In that regard, it is appropriate to recall that the Court of Justice has held that, in order to ascertain whether conduct was capable of producing foreclosure effects, the Commission was required inter alia to assess the possible existence of a strategy aiming to exclude competitors at least as efficient (see, to that effect, judgment of 6 September 2017, Intel v Commission, C‑413/14 P, EU:C:2017:632, paragraph 139). However, contrary to what Google claims, it cannot be inferred from this that the Commission must systematically demonstrate the existence of such a strategy in order to find an infringement of Article 102 TFEU.

679    While the existence of any anticompetitive intent constitutes one of a number of facts which may be taken into account in order to determine that a dominant position has been abused, it must be recalled that the Commission is under no obligation to establish the existence of such intent on the part of the dominant undertaking in order to render Article 102 TFEU applicable (see, to that effect, judgments of 19 April 2012, Tomra Systems and Others v Commission, C‑549/10 P, EU:C:2012:221, paragraphs 20 and 21, and of 30 January 2020, Generics (UK) and Others, C‑307/18, EU:C:2020:52, paragraph 162).

680    It follows that the circumstance, alleged by Google, that it did not intend to exclude a competitor as efficient as it is incapable of calling into question the foreclosure effects found in the contested decision.

681    Consequently, the Commission cannot be criticised, on the one hand, for not having established that Google had adopted a strategy aimed at excluding competitors at least as efficient as it and, on the other hand, for not having taken into account the fact that Google had not intended to exclude such competitors.

(v)    Conclusion on the as-efficient competitor test

682    Having regard to the foregoing, it is appropriate to find that Google is not justified in arguing that the Commission did not establish that a competitor as efficient as it could not emerge on the market for online search advertising intermediation or that such a competitor was capable of being excluded from that market, notwithstanding the evidence put forward by it and the lack of a strategy implemented by it in that regard.

(4)    Duration of GSAs and the unilateral termination right of certain direct partners

683    In the contested decision, the Commission found that all sites direct partners had been subject to the exclusivity clause over a long period, from 1 year to over 10 years. It also found, in essence, that the fact that the duration of GSAs concluded with the said direct partners, taken individually, was short, that is to say, generally less than two years, was irrelevant because many of those GSAs had been extended, sometimes several times, without substantial modifications. Last, it noted that the unilateral termination right of a direct partner did not prevent the application of the exclusivity clause until such time as that direct partner exercised that right.

684    Google claims that the various intermediaries were able to compete on the merits when GSAs were negotiated or renegotiated or where a unilateral termination right could be exercised. Consequently, it criticises the Commission, on the one hand, for having confused the overall length of the commercial relationship with all sites direct partners with the duration of each of the GSAs concluded with those direct partners and, on the other hand, for not having taken unilateral termination rights into account.

685    In that regard, Google states that most GSAs had a term of two years or less, such that, for example, in 2011, GSAs representing [confidential]% of the total turnover generated by AFS in the EEA had to be renewed in the following two years. In addition, it notes that almost a third of the said direct partners had negotiated unilateral termination rights. Last, it recalls that direct partners could, at any time, opt to use a competing online search advertising intermediation service on websites which were not included in their GSAs.

686    Surfboard claims that the Commission failed to consider the fact that, over the eight years in which it used AFS, four different GSAs were concluded and at least 10 renewals or amendments – corresponding to more than one amendment per year on average – were achieved. It states, in that regard, that those different negotiations gave it opportunities to terminate its GSAs and to secure many benefits from Google through competition.

687    The Commission contends that Google is wrong to claim that the other intermediaries could compete with it when GSAs were being renewed. First, the claim that, in 2011, GSAs representing [confidential]% of AFS’ turnover in the EEA had to be renewed in the following two years is irrelevant in so far as that claim concerns all GSAs and not only those concluded by all sites direct partners. Second, it is apparent from Annex A.46 to the application that the other intermediaries did not, in fact, have the possibility of competing with Google when renewing the GSAs concluded by 29 of the 34 all sites direct partners identified in the contested decision. Either those GSAs were extended ‘before they came up for renewal’ or the direct partners concerned concluded only one GSA which thus never ‘came up for renewal’. Third, on the one hand, it is irrelevant whether direct partners had a unilateral termination right since the exclusivity clause continued to apply until that right was exercised. On the other hand and in any event, it is apparent from Annex A.46 to the application that none of the all sites direct partners could exercise such a right at any moment.

688    Furthermore, the Commission submits that Google complains – for the first time in the reply – that the contested decision ‘fails to examine the duration of all GSAs’ containing the exclusivity clause, meaning that that argument is belated and, accordingly, inadmissible.

689    As far as Surfboard is concerned, the Commission notes that that company did not amend its GSA during the period between 1 July 2007 and 30 June 2010 mentioned by the contested decision. In addition, Surfboard had no right to terminate at any moment during that same period.

(i)    Admissibility of Google’s line of argument

690    According to Article 84(1) of the Rules of Procedure of the General Court, no new plea in law may be introduced in the course of proceedings unless it is based on matters of law or fact which have come to light in the course of the procedure.

691    However, a plea which constitutes an amplification of a plea previously made, either expressly or by implication, in the original application and is closely linked to it must be declared admissible. To be regarded as an amplification of a plea or a head of claim previously advanced, a new line of argumentation must, in relation to the pleas or heads of claim initially set out in the application, present a sufficiently close connection with the pleas or heads of claim initially put forward in order to be considered as forming part of the normal evolution of debate in proceedings before the Court (judgment of 8 July 2020, VQ v ECB, T‑203/18, EU:T:2020:313, paragraph 56).

692    In the case at hand, it should be noted that, in paragraph 86 of the application, Google criticises the Commission for not having taken account of the fact that competing intermediaries were able to compete with it for all or part of the demand of an all sites direct partner during the negotiation or renegotiation of a GSA or where a unilateral termination right existed. Additionally, in paragraph 99 of the application and in Annex A.46 thereto, Google claims that the Commission should have taken into account the duration of each GSA rather than the length of its commercial relationship with the all sites direct partners.

693    Consequently, and having regard to the case-law recalled in paragraph 691 above, it should be noted that Google’s line of argument, developed in the reply and seeking to criticise the Commission for not having examined the duration of each of the GSAs at issue, presents a sufficiently close connection with that developed in the application in order to be considered as forming part of the normal evolution of debate in proceedings before the Court, such that it cannot be regarded as being out of time. Accordingly, contrary to what the Commission contends, that line of argument must be deemed admissible.

(ii) Merits of Google’s line of argument

694    It should be recalled that, as has been noted in paragraph 390 above, the Commission had to take into account all the relevant circumstances in order to determine whether conduct was in fact capable of producing a foreclosure effect vis-à-vis a competitor at least as efficient as Google.

695    Regarding, more specifically, an exclusive supply obligation, it should be stated that the duration of that obligation, whether or not it is undertaken in consideration of the grant of a rebate, falls within the scope of such circumstances (see, to that effect, judgments of 26 January 2022, Intel Corporation v Commission, T‑286/09 RENV, under appeal, EU:T:2022:19, paragraph 507, and of 15 June 2022, Qualcomm v Commission (Qualcomm – exclusivity payments), T‑235/18, EU:T:2022:358, paragraph 425).

696    Indeed, as is emphasised in paragraph 36 of the Communication from the Commission entitled ‘Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings’, in general, the longer the duration of the obligation, the greater the likely foreclosure effect, it being specified that, if the dominant undertaking is an unavoidable trading partner for all or most customers, even an exclusive purchasing obligation of short duration can lead to anticompetitive foreclosure.

697    Moreover, it is appropriate to take into account, for the purposes of the assessment of the duration of an exclusive supply obligation, the economic and legal context of that obligation. In that regard, it is inter alia necessary to take into consideration the nature of the goods or services affected by that obligation as well as the actual conditions of the functioning and structure of the market or markets in question.

698    In the case at hand, first, it is apparent from recitals 388 and 398 of the contested decision, and from Table 15 set out therein, that the Commission is limited to determining the total period during which each of the all sites direct partners had been party to a GSA typically containing all of its websites in order to determine the period during which those direct partners had been obliged to source all or most of their requirements from Google. It thus took into account only the cumulative duration of the GSAs in which those partners had typically included all of their websites. Second, in recital 399 of that decision, the Commission noted that the unilateral termination right of a direct partner did not prevent the application of the exclusivity clause until such time as that direct partner exercised that right.

699    Thus, it is apparent from the contested decision and from Annex A.46 to the application, the content of which the Commission does not dispute, that that institution did not take into account the duration of each of the GSAs concluded with the all sites direct partners, taken individually, or the duration of each of the possible extensions of those GSAs. Nor did it take into account the actual conditions and the terms under which those extensions had been agreed or the substance of the clauses providing for the unilateral termination rights held by some of the all sites direct partners or the conditions in which those rights could be exercised.

700    In those circumstances, even assuming that all of the all sites direct partners identified in the contested decision had typically included all of their websites in their GSAs, the Commission could not, solely on the basis of the considerations recalled in paragraph 698 above and without having examined the actual conditions and the terms under which the extensions of the GSAs had been agreed, as well as the substance of the clauses providing for the unilateral termination rights held by some of the all sites direct partners and the conditions in which those rights could be exercised, exclude that those direct partners had the option of sourcing from Google’s competing intermediaries at the term of each of their GSAs, including before any extension of them, or before a unilateral termination right had been exercised. It follows that, in those conditions, the Commission likewise could not hold that the said direct partners had been obliged to source all or most of their requirements from Google for the entire cumulative duration of their GSAs in such a way that the said intermediaries had not had the possibility of disputing the share of the said market covered by those GSAs for that duration.

701    That conclusion is not called into question by the Commission’s arguments.

702    In the first place, contrary to what that institution essentially argues, the examination of the cumulative duration of the GSAs in which all sites direct partners had typically included all of their websites, as is presented in the contested decision, was not sufficient on its own to demonstrate that the exclusivity clause in those GSAs had produced a foreclosure effect.

703    In that regard, it should be noted that Google had argued, during the administrative procedure, that, having regard to the short duration of GSAs, those accounting for [confidential]% of the total AFS turnover from direct partners in the EEA had to be renewed between May 2011 and May 2013. It is true that the Commission calls into question the relevance of that item of data, in so far as it concerns all direct partners in the EEA and not only all sites direct partners. However, it should be noted that it assessed, in the contested decision, the capability of Google’s competing intermediaries to access the market for online search advertising intermediation taken as a whole. The fact – which is undisputed – that the GSAs accounting for [confidential]% of the total AFS turnover from direct partners in the EEA had to be renewed between May 2011 and May 2013 was a circumstance relevant to the examination of the capability of the exclusivity clause in GSAs concluded with all sites direct partners to produce a foreclosure effect, which was not taken into account by the Commission.

704    In the second place, the Commission contends, in its defence, that 29 of the 34 all sites direct partners identified in the contested decision had concluded only one GSA which either had not been extended or had been extended before its term. It infers from this that Google’s competitors had not been able to ‘contest that demand’.

705    However, it is appropriate to note that such a line of argument does not appear in the contested decision. It must be recalled that, as follows from paragraph 441 above, the Court cannot substitute its own reasoning for that of the Commission and the latter cannot supplement the statement of reasons for the contested decision during the proceedings.

706    In any event, regarding certain of the direct partners identified in the defence, it should indeed be noted, as the Commission does, that they had concluded only one GSA – which was not extended – in which they had typically included all of their websites. However, the Commission did not find, in the contested decision, that the exclusivity clause contained only in the GSAs concluded with those direct partners had been capable of restricting competition. In addition, it must be pointed out that those GSAs, in so far as they included, according to the Commission, typically all of the websites of the direct partners at issue, had a duration of between one year and three years and three months. Without it being necessary to rule on the question of whether the duration of those GSAs was sufficient, in this case, to allow the exclusivity clause in those GSAs to be capable of producing a foreclosure effect, suffice it to note that the said GSAs did not support the assertion, contained in recital 388 of the contested decision, that direct partners had been under an obligation to source all or most of their requirements for a duration that could have exceeded 10 years. Last, certain of those GSAs provided for unilateral termination rights in favour of the direct partners concerned, in respect of which the Commission examined neither the substance of the clauses providing for those rights nor the actual conditions in which they could be exercised, as has been noted in paragraph 699 above.

707    As for the other direct partners identified in the defence, the Commission notes that some, if not all, of the GSAs concluded by those direct partners in which they had typically included all of their websites had been extended before their term. However, contrary to what the Commission contends, it cannot necessarily be inferred from this that the other intermediaries could not compete with Google before that term. Indeed, as Google argues, the Commission seems to rely on the unsubstantiated premiss according to which the said intermediaries were able to compete with Google only when GSAs were renewed, that is to say, when new GSAs were signed, but not when existing GSAs were extended. More specifically, it is appropriate to point out that the Commission puts forward no element to show that the negotiations on the extension of a GSA could not be done at the end of a competitive process by which the direct partner concerned compared the services provided by Google and by the latter’s competitors.

708    Moreover, in response to a question from the Court during the hearing, the Commission added that Google’s competing intermediaries did not necessarily know the expiry dates of the GSAs of direct partners and that they had in any event assumed that GSAs would have been extended before their term, such that they had not been able to offer their services to the said direct partners.

709    However, other than the fact that those explanations do not appear in the contested decision, it should be noted that the Commission does not identify the reason for which direct partners could neither communicate the date of expiry of their GSAs to Google’s competing intermediaries nor engage in negotiations with them before deciding whether or not to extend them.

710    In the third place, it should be pointed out that the Commission refers to the judgment of 1 April 1993, BPB Industries and British Gypsum v Commission (T‑65/89, EU:T:1993:31, paragraph 73), cited in footnote 571 to the contested decision, to find that it did not have to take into account the unilateral termination rights held by some of the all sites direct partners, in so far as the exclusivity clause continued to apply until those rights had been exercised.

711    However, it is necessary to recall that, in the judgment of 1 April 1993, BPB Industries and British Gypsum v Commission (T‑65/89, EU:T:1993:31, paragraph 73), the Court had held that the entitlement of customers to discontinue their contractual relations with the undertaking at issue that was dominant at the time was, in essence, irrelevant, since the ‘legal possibility of termination’ was ‘illusory’. In the case at hand, the Commission indicated neither in the contested decision nor in response to the Court’s questions during the hearing that it regarded the exercise of unilateral termination rights by direct partners as illusory.

712    Nor has the Commission contended that Google was an unavoidable partner for all or most of the direct partners, within the meaning of paragraph 36 of the Communication from the Commission entitled ‘Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings’.

713    It follows that the Commission had to take into account the unilateral termination rights held by some of the all sites direct partners in order to determine whether Google’s competing intermediaries could access the part of the market for online search advertising intermediation that was covered by the exclusivity clause in GSAs concluded with those direct partners for the duration of those GSAs.

714    That conclusion is not called into question by the fact, put forward by the Commission in the defence, that Google identifies only 8 direct partners of the 34 all sites direct partners identified in the contested decision that had held a unilateral termination right and that those rights could not be exercised at any moment. Indeed, as Google argues and as follows from paragraphs 699 and 700 above, the Commission ought to have examined the substance of the clauses providing for those rights and the conditions in which they could be exercised, but also which direct partners could enjoy them, in order to determine whether the said rights were capable of calling into question, at least to a certain extent, the fact, found in the contested decision, that the exclusivity clause in GSAs concluded with all sites direct partners had prevented Google’s competitors from accessing a significant part of the market for online search advertising intermediation for the duration of those GSAs.

715    It is apparent from the foregoing that, even though the Commission cannot be criticised for having taken into account, in the contested decision, the cumulative duration of the various GSAs in which direct partners had typically included all of their websites, as one of the relevant circumstances of the case, it is appropriate to hold that it should have ascertained whether, in view of the legal and economic context of the case, all sites direct partners had the option of sourcing from Google’s competing intermediaries at the term of each of their GSAs, including before any extension of them, or before a unilateral termination right had been exercised.

(5)    Conclusion on the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

716    As has been found in paragraph 650 above, the Commission was right to consider that, having regard to the coverage of the placement clause in the light of the circumstances recalled in paragraph 602 above, the coverage of the exclusivity clause in GSAs in which direct partners had typically included all of their websites could be sufficient to enable that clause to be capable of producing a foreclosure effect between 1 January 2006 and 31 December 2015. In addition, it has been found in paragraph 682 above that Google was not justified in arguing that the Commission had failed to prove that a competitor as efficient as it could not emerge on the market for online search advertising intermediation or that such a competitor was capable of being excluded from that market.

717    However, on the one hand, as follows from paragraph 650 above, the Commission did not establish that the exclusivity clause could have produced a foreclosure effect, owing to its coverage, between 1 January 2016 and 31 March 2016. On the other hand, as follows from paragraph 715 above, it failed to take into consideration all the relevant circumstances of the case in the context of the assessment of the duration for which all sites direct partners were obliged, by virtue of that clause, to source all or most of their requirements exclusively from Google, even assuming that all of the all sites direct partners identified in the contested decision had been under such an obligation.

718    It follows that, contrary to what the case-law recalled in paragraph 107 above requires, the Commission has not demonstrated to the requisite legal standard, in the light of all the circumstances of the case, that the exclusivity clause in GSAs concluded with all sites direct partners was capable of preventing Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA, for the duration for which that clause applied.

(c)    Conclusion on the third part of the second plea

719    As has been recalled in paragraph 392 above, in recital 362 of the contested decision, the Commission considered that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition, in the light of all the circumstances relevant to the case. It noted in that regard that that clause had (i) deterred those direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers.

720    As has been recalled in paragraphs 393 and 394 above, the Commission essentially considered that the exclusivity clause in GSAs concluded with all sites direct partners was capable of producing a foreclosure effect, finding that it had, on the one hand, deterred those direct partners from sourcing from Google’s competing intermediaries and had, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation. In addition, it inferred from that foreclosure effect that the said clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

721    As has been recalled in paragraph 399 above, in recital 364 of the contested decision, before examining the effects of each of the five restrictions of competition that it identified (see paragraph 719 above), the Commission specified that, for the purposes of its analysis seeking to demonstrate that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition, it had taken into account the ‘duration of [that c]lause’, referring in that regard to Section 8.3.4.2 of that decision, relating to the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation, and thereby highlighting, rightly, the importance of the developments set out in recitals 388, 398 and 399 appearing in that section. It also indicated that it had taken into consideration the coverage rate of the said clause, which it examined in the same section of that decision. It is also apparent from the systemic place of the said recital 364 in the structure of the decision in question that the Commission took into account the duration and coverage rate of the exclusivity clause when it examined its effects in the context of each of the five restrictions identified in the said decision.

722    As has been noted in paragraph 717 above, however, the Commission failed to take into consideration all the relevant circumstances of the case in the context of the assessment of the duration for which it considered that, by virtue of the exclusivity clause, all sites direct partners had been obliged to source all or most of their requirements from Google.

723    What is more, as has also been found in paragraph 717 above, the Commission did not establish that the exclusivity clause in GSAs concluded by all sites direct partners could have produced a foreclosure effect, owing to its coverage, between 1 January and 31 March 2016.

724    It follows that the errors committed by the Commission, recalled in paragraphs 722 and 723 above, vitiate all of the restrictions identified by it in the contested decision, such that it is appropriate to conclude that the Commission has not demonstrated, to the requisite legal standard, that the exclusivity clause in GSAs concluded with all sites direct partners had been capable of deterring those direct partners from sourcing from Google’s competing intermediaries or that it had been capable of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA and, consequently, that that same clause had been capable of having the foreclosure effect found in that decision.

725    In those conditions, it is appropriate to find that the Commission has also not demonstrated, to the requisite legal standard, that the exclusivity clause in GSAs concluded with all sites direct partners had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

726    Moreover, it must be pointed out that, in recitals 422 and 423 of the contested decision, the Commission essentially found that the English clause had exacerbated the foreclosure effect of the exclusivity clause in GSAs concluded with all sites direct partners by further deterring those direct partners from sourcing from Google’s competing intermediaries. However, the Commission has neither established that the exclusivity clause in those GSAs was capable of having such an effect nor alleged that the English clause alone was capable of having that effect. Accordingly, the English clause cannot suffice, on its own, to demonstrate that the exclusivity clause in the said GSAs constituted an infringement of Article 102 TFEU.

727    It follows that the third part of the second plea must be upheld, without it being necessary to rule on the other arguments of Google under that plea, and, consequently, that the contested decision must be annulled to the extent that it found that the exclusivity clause in GSAs concluded with all sites direct partners constituted an infringement of Article 102 TFEU.

D.      Third plea: the placement clause did not constitute an abuse of a dominant position

728    By the third plea, Google criticises the Commission for having found that the placement clause constituted an abuse of a dominant position within the meaning of Article 102 TFEU. This plea consists of two parts, the first alleging that the scope of that clause was misinterpreted, and the second alleging that there was no restriction of competition.

729    As a preliminary point, it must be recalled that, in the March 2009 template GSA, the placement clause was worded as follows:

‘The parties agree that: … if Google is providing its AFS service to Company under an Agreement in relation to any Site(s), Company shall at all times during the Term request at least three (3) wide format AFS Ads from Google in relation to each Search Query submitted on such Site(s) and shall display the AFS Ads provided by Google on the applicable Results Pages such that (i) no Equivalent Ads appear above or directly adjacent to such AFS Ads, and (ii) the AFS Ads are displayed in a single continuous block and are not interspersed with other advertisements or content.’

730    Clause 1.1 of the March 2009 template GSA stipulated that the expression ‘Equivalent Ad’ had to be understood as referring to ‘any advertisements that are the same as or substantially similar in nature to the AFS Ads provided by Google under any Agreement’.

731    Between June 2010 and October 2013, the placement clause of the template GSA was amended, so far as concerns mobile and tablet ads, in the following manner:

‘The parties agree that: if Google is providing its AFS service to Company under an Agreement in relation to any Site(s), Company shall at all times during the Term request …, if the AFS Request has been generated by a Search Query submitted by an End User using a Mobile Device or Tablet Device, at least one (1) Mobile Search Ad or at least one (1) Tablet Search Ad, as applicable …’

1.      First part of the third plea: misinterpretation of the scope of the placement clause

732    In the contested decision, the Commission considered that the placement clause reserved the most prominent space on the websites of direct partners for Google search ads.

733    First, the Commission found that direct partners could not show competing ads ‘above or immediately next to’ those of Google. It inferred that, in principle, Google ads were shown at the top-left position of the search results page, above the search results. Moreover, where no ad was shown at the top-left position of the said page, it noted that Google ads had to be shown in the space that the user viewed first, which could be at the bottom of the page.

734    Second, the Commission observed that Google, on the one hand, referred to the placement clause as a ‘relaxed exclusivity’ clause and, on the other hand, included that clause in the section entitled ‘exclusivity’ of certain GSAs. It also quoted Google’s replies to requests of certain direct partners in which it was inter alia stated that Google did not wish for competing ads to be shown in a ‘more favourable’ or ‘better’ position than its own ads.

735    Third, the Commission considered that the space above the search results was the most profitable space and that the space allocated to competing ads was therefore less profitable.

(a)    Possibility of showing competing ads below Google ads

736    Google claims that the Commission was wrong to consider that the placement clause prohibited direct partners from showing competing ads below its own ads. It notes that the expression ‘directly adjacent’ could not be interpreted as meaning ‘vertically adjacent to Google search ads’ since it had been specified, in the template GSA, that the said clause applied to competing ads shown ‘above’ its own ones or ‘directly adjacent’ to them. After all, if the word ‘adjacent’ referred to ads shown in a ‘vertically adjacent’ manner, it would not have been necessary to refer to ads shown ‘above’. Moreover, the contested decision does not identify any direct partner which interpreted the word ‘adjacent’ in the same way as the Commission had.

737    Vinden and Surfboard submit that they also interpreted the placement clause as meaning that it authorised the display of competing ads below Google ads.

738    The Commission contends that Google does not dispute that it follows from the ordinary meaning of the words ‘directly’ and ‘adjacent’, used in the placement clause, that direct partners interpreted that clause as prohibiting the display of competing ads vertically adjacent to its own ads. In addition, a number of GSAs indicated that the prohibition on displaying competing ads ‘directly adjacent’ to Google ads was to be interpreted as meaning that competing ads could not be displayed either ‘below’ them or ‘directly adjacent’ to them. Last, Google identified, during the administrative procedure, only a single direct partner that had interpreted the said clause as authorising competing ads also to be displayed below its own ads.

739    As far as Vinden and Surfboard’s line of argument is concerned, the Commission adds that they have not demonstrated that the placement clause allowed the display of ads immediately below Google’s ones. Specifically, it notes that the evidence put forward by Vinden dates from 2020, meaning that it concerns a period in which that clause no longer applied. It is also apparent from that evidence that Vinden showed competing ads only below search results.

740    In the first place, it should be noted that the Commission defined the scope of the placement clause in different ways in the contested decision. First, as follows from paragraph 733 above, the Commission was limited to indicating, in recital 465 of that decision, that the placement clause prohibited direct partners from showing competing ads ‘above or immediately next to’ Google ads. Second, in recital 481 of that decision, in the part of its reasoning on the response to the arguments developed by Google during the administrative procedure, it stated that the phrase ‘directly adjacent’ also referred to competing ads displayed immediately below Google ads.

741    In the second place, it should be borne in mind that, in considering in the contested decision that the English word ‘adjacent’ also referred to competing ads placed below Google ads, the Commission relied on two definitions of that word contained in the online dictionaries Oxford English Dictionary and Merriam Webster. It is apparent from those definitions that the latter word means, on the one hand, ‘next to or adjoining something else’ and, on the other hand, ‘not distant’, ‘having a common endpoint or border’ and ‘immediately preceding or following’.

742    It follows that, while it is true that the word ‘adjacent’ can refer to anything that can surround something, it follows from the definitions cited in the contested decision that that word can also specifically designate what is ‘next to’ that thing. Consequently, it should be considered, as Google does, that the exact meaning of the word in question depends on the context in which it is used and that it does not necessarily designate what is below the said thing.

743    In that regard, first, it should be noted that the Commission identified, in the contested decision, eight GSAs in which the expressions ‘directly adjacent’ was defined as referring to competing ads displayed ‘below and adjacent’.

744    However, on the one hand, it follows from Annex A.53 to the application, the content of which is not disputed by the Commission, that, contrary to what that institution maintains, the clauses of the eight GSAs mentioned in the contested decision did not define the terms ‘directly adjacent’ as meaning ‘below and adjacent’. Additionally, it is appropriate to note, as Google argues, that those clauses specifically stated that competing ads could not be shown ‘below’ Google ads or ‘adjacent’ to them. Thus, the wording of the said clauses suggests that the word ‘adjacent’ did not suffice, on its own, to refer to the competing ads that would have been displayed ‘below’ Google ads.

745    On the other hand, it should be noted that all of the eight GSAs mentioned in the contested decision were concluded with the [confidential] group alone. In those conditions, contrary to what the Commission contends, it cannot in any event be inferred from the specific wording of those eight GSAs, agreed with that direct partner in particular, that the words ‘directly adjacent’, assuming that they were used in other clauses of those GSAs, had the same meaning in those agreements as in those concluded with the other direct partners.

746    Second, it should be noted, as Google does, that the fact that the template GSA, partly reproduced in recital 91 of the contested decision, stipulates that competing ads could not be displayed ‘above’ Google ads or ‘directly adjacent’ to them also implies that the words ‘directly adjacent’ did not suffice, on their own, to refer to competing ads that would have been displayed ‘above’ Google ads.

747    In the third place, it should be noted that the Commission acknowledged, in the contested decision, that one of the direct partners had interpreted the placement clause as meaning that it authorised the display of competing ads below Google ads. However, contrary to what the Commission suggests, the mere fact that Google had identified in that regard, during the administrative procedure, only one direct partner cannot mean that the other direct partners would have interpreted that clause differently. First, it cannot be inferred from the choice of direct partners not to display competing ads below Google ads that that choice followed exclusively from their interpretation of the said clause, to the exclusion of all other considerations. Second, it should be noted, as Google does, that, despite the investigatory powers conferred on it by Regulation No 1/2003, the Commission identified no direct partner that had interpreted that clause as meaning that it prohibited such a display.

748    In those conditions, it is appropriate to find that the Commission has not established that the placement clause prohibited direct partners from displaying competing ads below Google ads.

(b)    Spaces generating the highest CTR

749    Google claims that the placement clause did not require direct partners to show its own ads in the space generating the highest CTR. On the one hand, it asserts that competing ads could, in certain configurations, be shown in spaces generating higher CTR than those reserved for its own ads. In that regard, it states that, while the Commission considered, in the contested decision, that publishers ‘rarely’ adopted those configurations, the latter were nevertheless not prohibited by the said clause. On the other hand, it asserts that direct partners could display competing ads above search results and that, in those conditions, the CTRs of those ads were comparable with those of its own ads.

750    Surfboard submits that the placement clause allowed it to display competing ads in spaces leading to CTRs comparable to those of Google ads.

751    The Commission disputes Google and Surfboard’s line of argument.

752    In that regard, it is appropriate to examine, first, the illustrations reproduced in Annex A.52 to the application and, second, the screenshots of results pages of direct partners corresponding to Figures 5 and 6 of Annex C.11 to the reply.

(1)    Illustrations in Annex A.52 to the application

753    It must be noted that Annex A.52 to the application reproduces the two following illustrations of configurations, in compliance with the placement clause, which enabled competing ads to generate CTRs higher or comparable to those of Google ads:

754    In the first place, it should be noted that it follows from Tables 18 to 22 of the contested decision that ads displayed under search results generated CTRs higher than those of ads displayed to the right of those results between 2012 and 2015. Consequently, it is appropriate to consider, as Google does, that, in the configuration corresponding to Figure 1, competing ads could generate higher CTRs than its own ads, which the Commission moreover does not dispute.

755    However, the Commission noted, in the contested decision, that that configuration was rarely chosen by direct partners, which Google likewise does not dispute. While it does not expressly set out the reasons for that situation, it nevertheless follows from Tables 18 to 22 of that decision that the ads shown above search results generated the highest CTRs. Thus, by showing no ads above those results, direct partners accepted a reduction in the number of clicks generated by their results pages and, therefore, in their revenues. Consequently, direct partners had to accept to limit their total revenues in order to allow competing ads to generate higher CTRs than Google ads.

756    In those conditions, it is appropriate to find that, in the light of the circumstances of the case, Figure 1 does not illustrate a configuration which permitted, in practice, direct partners to allocate to competing ads spaces generating CTRs higher than those reserved for Google ads.

757    In the second place, it should be noted that, as follows from paragraph 748 above, direct partners could display competing ads below Google ads. It follows that Figure 2 must be deemed to represent a configuration compliant with the placement clause.

758    However, it is settled that direct partners had to display at least three Google ads within the same ‘block’ when the online query was made on a desktop device.

759    On the one hand, it must be noted that, contrary to what Google claims, it does not follow from Tables 18 to 22 of the contested decision that the ads displayed in the fourth slot above search results generated a CTR only slightly lower than those of ads displayed in the first three slots. Specifically, while it is true that the difference in CTRs between the ads displayed in the third and fourth slots is not significant – incidentally, in 2015, the CTR of ads displayed in the fourth slot was even higher than that of ads displayed in the third slot – it should be noted, as the Commission does, that there is a significant discrepancy between the CTRs of ads displayed in the first slot and the CTRs of those displayed in the fourth slot. Thus, for example, for 2013, it is apparent from the table below that the CTR of ads displayed in the fourth slot was [confidential]%, whereas that of ads displayed in the first slot was [confidential]%. It follows that, in 2013, ads displayed in the first slot generated [confidential]% more clicks than ads displayed in the fourth slot. Similarly, in 2015, the CTR of ads displayed in the fourth slot was [confidential]%, whereas that of ads displayed in the first slot was [confidential]%. It follows that, in 2015, ads displayed in the first slot generated [confidential]% more clicks than ads displayed in the fourth slot:

760    On the other hand, it follows in particular from the third column of Table 23 of the contested decision that the difference in CTRs between the first three ads displayed and those which follow below (but above the search results) generally increases with the number of ads displayed. Thus, the act of displaying a larger number of competing ads does not call into question the fact that, in Figure 2, Google ads occupied the spaces generating the highest CTRs.

761    Consequently, it is appropriate to consider that Google is not justified in claiming that the configuration corresponding to Figure 2 enabled competing ads to generate CTRs comparable to those of its own ads.

(2)    Figures 5 and 6 of Annex C.11 to the reply

762    Google claims that direct partners could display competing ads adjacently to its own ads, which could thus generate CTRs comparable to those of its own ads. It refers in that regard to the screenshots of websites of two direct partners which are reproduced in Figures 5 and 6 of Annex C.11 to the reply.

763    However, in the first place, it should be noted that Google’s line of argument tends to call into question the very terms of the placement clause prohibiting competing ads from being displayed adjacently to its own ads. Nevertheless, the fact that two direct partners failed to observe those terms, at a given moment, does not suffice to prove that that clause permitted such a display.

764    In the second place, it is appropriate to note that Google acknowledged, during the hearing, first, that direct partners could not display competing ads above its own ads and, second, that Figures 5 and 6 of Annex C.11 to the reply showed competing ads that were not displayed above the search results, as the Commission had observed in footnote 81 to the rejoinder. It follows that it was competing ads that must be regarded as being positioned on the right-hand side of the results page. It follows from Tables 18 to 22 of the contested decision, the content of which is not disputed by Google, however, that such ads generated a significantly lower CTR than that of ads situated at the top of a results page.

765    Consequently, it is appropriate to find that, even assuming that Figures 5 and 6 of Annex C.11 to the reply show competing ads displayed in compliance with the placement clause, such a display did not enable the generation, for those ads, of CTRs comparable to those of Google ads.

766    In those conditions, it must be found that Google is not justified in claiming that direct partners had been able, in accordance with the placement clause, to adopt configurations in which competing ads generated CTRs higher or comparable to those of its own ads.

(c)    Conclusion on the first part of the third plea

767    Google has not proved that, in practice, the CTR of competing ads could be at least comparable with that of its own ads, as displayed in accordance with the placement clause. It follows that, as regards online search ads, the Commission was right to find that the placement clause reserved the most prominent spaces of the results pages of direct partners for Google ads.

768    Accordingly, the first part of Google’s third plea must be rejected as unfounded.

2.      Second part of the third plea: no restriction of competition as a result of the placement clause

769    In the contested decision, the Commission considered, in recital 494 thereof, that the placement clause was capable of restricting competition, in the light of all the circumstances of the case. In that regard, it noted that that clause had (i) deterred direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers. In addition, it found that the binding nature of the mock-ups had exacerbated the capability of the said clause to restrict competition.

770    More specifically, it should be noted that, in finding that the placement clause had, on the one hand, deterred direct partners from sourcing from Google’s competing intermediaries and, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, the Commission essentially considered that that clause was capable of producing a foreclosure effect.

771    In addition, it should be noted that the Commission inferred from the foreclosure effect of the placement clause that that clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

772    First, it follows from recitals 530 to 532 of the contested decision that the foreclosure effect of the placement clause had deterred Google’s competing intermediaries from providing or developing different online search ads, with the result that that clause had deterred them from investing in innovation. Next, it follows from recital 534 of that decision that the foreclosure effect had deprived the said intermediaries of revenues and data that they could have used to provide online search ads. Last, it follows from recital 539 of the said decision that the foreclosure effect had allowed Google to set the prices paid by advertisers at a high level, thereby increasing the prices consumers paid for the goods featured in the online search ads. The Commission added, in recital 540 of the same decision, that the fact that the said clause had possibly deterred innovation had also deprived consumers of a wider choice of online search ads.

773    Google submits that the placement clause, first, did not produce the foreclosure effects found in the contested decision; second, did not help it to maintain or strengthen its dominant position on the national markets for online search advertising at issue; and, third, neither deterred innovation nor harmed consumers. In addition, it takes issue with the Commission for not having shown that the mock-ups were capable of restricting competition.

774    Surfboard and Vinden claim that the placement clause had no effect on their conduct. Vinden submits, moreover, that the mock-ups could be modified and denies that they were binding. Surfboard adds that the placement clause was objectively justified.

775    It is appropriate at the outset to examine the foreclosure effect identified in the contested decision resulting from the placement clause. It must therefore be ascertained whether that clause was capable, on the one hand, of deterring direct partners from sourcing from Google’s competing intermediaries and, on the other hand, of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation.

776    In that regard, it should be noted that, in recital 496 of the contested decision, the Commission specified that, for the purposes of its analysis seeking to demonstrate that the placement clause was capable of restricting competition, it had taken into account all the relevant circumstances, including, on the one hand, the extent of Google’s dominant position, both on the national markets for online search advertising at issue and on the market for online search advertising intermediation, and, on the other hand, the share of the latter market covered by the said clause and the ‘duration of [that c]lause’. It referred, in that regard, respectively, to Section 7, which includes the considerations recalled in paragraphs 401 to 404 above, and, in essence, to the whole of Section 8.4.4.2 of the said decision, relating to the impossibility for Google’s competing intermediaries to access a significant part of the said market.

777    In line with what has been found in paragraphs 400 and 405 above, it should be noted, on the one hand, that the approach followed by the Commission is in conformity with the case-law and, on the other hand, that Google does not challenge the content of Section 7 of the contested decision other than by arguing, under the first plea, that the Commission erroneously defined the relevant markets in Section 6 of that decision.

778    In addition, as has been stated in paragraph 767 above, the Commission was right to find that the placement clause reserved the most prominent spaces of the results pages of direct partners for Google ads. Consequently, it is appropriate to find that, as is apparent from recitals 335, 467, 630, point 2, 712 and 718 of the contested decision, the placement clause was akin to a relaxed exclusivity clause so far as concerns websites that were included in GSAs containing that clause.

779    As has been recalled in paragraph 406 above, although the ability of exclusivity clauses to exclude competitors is not automatic, as moreover is illustrated by paragraph 36 of the Communication from the Commission entitled ‘Guidance on the Commission’s enforcement priorities in applying Article [102 TFEU] to abusive exclusionary conduct by dominant undertakings’, the fact remains that, by reason of their nature, those clauses give rise to legitimate concerns of competition.

780    Taking into account those factors, it is appropriate to examine, as a first step, whether the placement clause was capable of deterring direct partners from sourcing from Google’s competing intermediaries and, as a second step, whether the said clause was capable of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation.

(a)    Deterrent effect of the placement clause vis-à-vis direct partners

781    In the contested decision, the Commission found that the placement clause had deterred direct partners from sourcing from Google’s competing intermediaries.

782    In that regard, first, the Commission found, in recital 499 of the contested decision, that the placement clause had prevented direct partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries. Second, it noted, in recital 500 of that decision, that that clause had prevented those direct partners from being able to adopt certain configurations for their results pages in so far as they by necessity had to display at least three Google ads in a single block when the user visited those websites from a desktop device. In addition, in response to an argument from Google, it considered, in recital 502 of that decision, that direct partners would have had a commercial interest in sourcing from Google’s competing intermediaries absent the placement clause.

(1)    Capability of the placement clause to prevent direct partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries

783    Google claims, first, that the placement clause did not prevent direct partners from evaluating the commercial impact of sourcing from competing intermediaries and, second, that the statements of two direct partners, set out in recital 499 of the contested decision, were not such as to establish that that clause had actually prevented direct partners from evaluating the commercial interest in sourcing from such intermediaries.

(i)    Scope of the placement clause

784    Google essentially claims that, in view of its scope, the placement clause was not such as to prevent direct partners from evaluating the commercial interest insourcing from Google’s competing intermediaries. Specifically, it argues that (i) direct partners were free to choose which websites to include in their GSAs, (ii) other intermediaries were able to compete with it when those GSAs were negotiated or renegotiated or where the said direct partners held a unilateral termination right, (iii) direct partners could use AFS by concluding online contracts, and (iv) the placement clause expressly allowed competing ads to be displayed.

785    Vinden claims that the placement clause did not deter it from displaying competing ads. It specifies that it also sourced online search ads from Yahoo! and that the GSA that it had concluded explicitly stated that it could display competing ads.

786    The Commission disputes Google and Vinden’s line of argument.

787    First, it is necessary to note that, as follows from paragraphs 412 to 415 above and as the Commission essentially observes, the fact that direct partners could freely experiment with competing ads on websites that were not included in their GSAs does not allow the effect of the placement clause to be assessed as regards websites that were in fact included in those GSAs. Thus, the fact that direct partners could choose the websites that they included in their GSAs, whether it be when negotiating or renegotiating them or indeed following the exercise of a unilateral termination right, or that they could choose to include their websites in an online contract is not such as to call into question the fact that that clause could have, for a certain duration, prevented them from evaluating the commercial interest in sourcing from Google’s competing intermediaries, at least as regards the websites included in their GSAs.

788    Second, it should be recalled that, as has been noted in paragraph 767 above, the Commission was right to find that the placement clause reserved, for the benefit of online search ads from Google, the most prominent spaces of the results pages of direct partners and thus the spaces generating the highest CTRs. Consequently, the circumstance that that clause allowed competing ads to be displayed does not call into question the fact that it necessarily limited the experiments that direct partners could carry out regarding the display of those ads.

789    In those conditions, it is appropriate to find that Google is not justified in claiming that the placement clause was not, in view of its scope, capable of preventing, at least to a certain degree, direct partners from evaluating the commercial interest in sourcing from competing intermediaries, particularly so far as concerns the websites included in GSAs containing that clause.

(ii) Statements of direct partners

790    Google notes that the Commission relied, in the contested decision, on the statements of only two direct partners to find that the placement clause had prevented direct partners as a whole from evaluating the commercial interest in sourcing from other intermediaries. In addition, it states that it follows from the statements of many direct partners – including one of the two mentioned in the contested decision – that that clause had not prevented them from evaluating that impact.

791    The Commission disputes the admissibility of Google’s line of argument aimed at calling into question the fact that direct partners had confirmed that the placement clause had prevented them from evaluating the commercial interest in sourcing from other intermediaries, on the ground that that line of argument was developed for the first time in the reply. In addition, it considers that that line of argument is, in any event, unfounded.

–       Admissibility of Google’s line of argument

792    It is true that it is at the stage of the reply that Google criticises the Commission for having relied only on the statements of two direct partners to find that the placement clause had prevented direct partners as a whole from evaluating the commercial interest in sourcing from competing intermediaries.

793    However, on the one hand, it should be noted that Google has expressly challenged, in paragraphs 125 and 126 of the application, the assertion, contained in recital 499 of the contested decision, according to which the placement clause had prevented direct partners from evaluating the commercial interest in sourcing from competing intermediaries.

794    On the other hand, it must be pointed out that Google specifies that its line of argument, developed in the reply, seeks to call into question paragraph 273 of the defence. In that regard, it should be noted that, according to that paragraph, the circumstance, put forward in the application by Google, that three other direct partners had been able to evaluate competing ads – particularly in order to evaluate the commercial interest in procuring more of them – on websites subject to the prior authorisation clause is irrelevant. It is settled that, as has been noted in paragraph 594 above, the prior authorisation clause applied only where the placement clause also applied. Consequently, the fact that direct partners, subject to the prior authorisation clause, could evaluate the commercial interest in sourcing from other intermediaries is also relevant to the examination of the effects of the placement clause.

795    In those conditions, it must be considered, having regard to the case-law recalled in paragraph 691 above, that Google’s line of argument presents a sufficiently close connection with the complaints initially set out in the application in order to be considered as forming part of the normal evolution of debate in proceedings before the Court and constitute an amplification of those complaints. Contrary to what the Commission contends, therefore, that line of argument cannot be rejected as inadmissible.

–       Merits of Google’s line of argument

796    As follows from the case-law recalled in paragraph 111 above, it is necessary, in principle, to attach a great evidential value to the replies to the Commission’s requests for information.

797    In the first place, it should be pointed out that, in recital 499 of the contested decision, the Commission states that a number of direct partners confirmed the fact that the placement clause had prevented them from evaluating the commercial interest in sourcing from Google’s competing intermediaries. It cites, in that regard, the replies of two direct partners, namely the [confidential] group and [confidential], to a request for information of 31 July 2015. Moreover, it follows from Table 26 of the contested decision that the Commission identified 53 direct partners that had concluded GSAs containing the placement clause.

798    In those conditions, it must be held that it is apparent from recital 499 of the contested decision that the Commission did not intend to compile an exhaustive list of all the replies that it received during the administrative procedure, but is limited to giving only examples of replies confirming the fact that the placement clause was capable of preventing direct partners from evaluating the commercial interest in sourcing from one of Google’s competing intermediaries.

799    Moreover, it should be noted that Google does not dispute the tenor of the reply of the [confidential] group, mentioned in recital 499 of the contested decision, according to which, in essence, the placement clause affected its advertising strategy, in particular by limiting the way in which it could display ads from Google’s competitors, which, in turn, prevented it from comparing Google ads and those of its competitors.

800    Furthermore, the Commission reproduced, in recital 499 of the contested decision, the extract of a reply of [confidential] indicating that it had wished to be able to put several different intermediaries in competition with each other in real time, for each online query, in order to determine which ad to display in the spaces normally reserved for Google ads under the placement clause. [confidential] explained, in essence, that the placement clause had potentially prevented it from enjoying an increase in turnover.

801    Google argues that [confidential] had also reported working with an undertaking other than Google on ‘sponsored links’. However, it should be noted that Google puts forward nothing enabling those links to be distinguished from the results of specialised searches, which it does not claim belong to the online search advertising market. In any event, it cannot be inferred from that fact that [confidential] had used the services of another intermediary in addition to AFS that, had it not been for the placement clause, it would not have wished to evaluate the commercial interest in displaying ads from that other intermediary in the most prominent spaces of its results pages.

802    In those conditions, it is appropriate to find that the Commission was right to consider that the reply of [confidential] was capable of corroborating the fact that the placement clause could have deterred direct partners from sourcing at least part of their requirements from Google’s other competing intermediaries.

803    In the second place, it is indeed appropriate to point out, as Google does, that certain other direct partners had indicated to the Commission that their advertising strategy had not been impacted by the placement clause.

804    First, it should be noted that, to the question of to what extent the clauses at issue had affected its advertising strategy, [confidential] had replied as follows: ‘Since [its] advertising strategy on its web properties considers Google AdSense for Search not its main business, those clauses did not prevent any other partnership with other intermediaries.’

805    In that regard, the Commission contends that [confidential]’s reply was irrelevant because it did not address the ability of that direct partner to evaluate competing ads.

806    However, on the one hand, in so far as both the question asked by the Commission and the reply formulated by [confidential] addressed the overall advertising strategy of that direct partner, it must be noted that that reply had necessarily dealt with the ability of that direct partner to evaluate the commercial interest in sourcing from intermediaries other than Google. Moreover, it is appropriate to note that the replies of the [confidential] group and of [confidential], quoted in recital 499 of the contested decision, also answered that same question. Thus, it is apparent from those replies that the [confidential] group and [confidential] also considered that the ability to evaluate the commercial interest in sourcing from such intermediaries was an element within the scope of their advertising strategy.

807    On the other hand, it is appropriate to find that [confidential]’s reply, in so far as it specified that the placement clause ‘[ha]d not prevent[ed] any other partnership with other intermediaries’, was such as to call into question the fact, found in the contested decision, that that clause had deterred that direct partner from sourcing from such intermediaries.

808    Accordingly, it should be noted that the Commission is not justified in arguing that [confidential]’s reply was irrelevant.

809    Second, it is appropriate to remark that the Commission had asked [confidential] to explain to what extent the placement clause had prevented it from displaying competing ads or had in practice limited its ability to display such ads. [confidential] had in that instance replied that it had not been ‘affected by th[e placement] clause and thus [had] never contested it, as [it had] decided that the inclusion of multiple equivalent third party search components would harm the user[’s] experience …’.

810    In that regard, the Commission contends that [confidential]’s reply was irrelevant because it specified that ‘the inclusion of multiple equivalent third party search components would harm the user experience’. However, it should be noted that the fact that a direct partner has explained the reason why it did not want to source from Google’s competing intermediaries is, on the contrary, a factor liable to lend credibility to the assertion that the placement clause had not deterred that direct partner from sourcing from such intermediaries.

811    Accordingly, it should be noted that the Commission is not justified in arguing that [confidential]’s reply was irrelevant.

812    Third, it should be borne in mind that [confidential] had indicated to the Commission, in reply to a request for information, that ‘no, [its] Partnership with Google d[id] not limit [its] ability to integrate other intermediaries’.

813    In that regard, the Commission argues that [confidential]’s reply was irrelevant because it did not concern the placement clause. However, it must first of all be borne in mind that the Commission does not put forward any evidence in support of its allegation or dispute the fact that [confidential] was a direct partner subject to a GSA. Next, it is apparent from Exhibits 6 and 7 of Annex C.15 to the reply that the question at issue concerned all of the contract clauses relating to AFS and that the said reply made express reference to the overall ‘partnership’ with Google. Last, it should be recalled that, as has been mentioned in paragraph 594 above, all GSAs containing the prior authorisation clause also contained the placement clause, meaning that, if that direct partner was subject to the prior authorisation clause, as the Commission seems to imply, it was also by necessity subject to the placement clause.

814    Accordingly, it should be noted that the Commission is not justified in arguing that [confidential]’s reply was irrelevant.

815    Fourth, it should be noted that the general counsel of [confidential], which is the parent company of [confidential], had, in a letter of 31 October 2016 submitted to the Commission as an annex to the response to the statement of objections, stated the following: ‘The premium placement clause has not had any effect on us. As noted above, we have moved users away from Google by directing our search traffic to [confidential], which is a domain that we’ve chosen to monetize using Yahoo!’s search ads instead of AFS’.

816    In that regard, the Commission contends that that letter was irrelevant because [confidential] had not claimed that competing ads generated a CTR comparable to that of Google ads. It adds that, in any event, the probative value of the said letter was limited because it does not know the ‘context’ in which Google had obtained it.

817    However, on the one hand, it must be borne in mind that there was no need for [confidential] to indicate that competing ads could generate a CTR comparable to that of Google ads to explain that the placement clause had not deterred it from sourcing from another intermediary. Moreover, none of the replies of the two direct partners quoted in recital 499 of the contested decision mentions the comparison of the CTRs between Google ads and those from other intermediaries.

818    On the other hand, it should be noted that the Commission does not claim that [confidential]’s letter was devoid of all probative value. Furthermore and in any event, the Commission could not, in the light of paragraphs 512 to 514 above, contest the relevance of [confidential]’s letter solely because it was submitted by Google, when it had been free to request further information from [confidential] pursuant to Article 18 of Regulation No 1/2003.

819    Accordingly, it should be noted that the Commission is not justified in arguing that [confidential]’s reply was irrelevant.

820    However, it must be stated that it is impossible to know from the other replies cited by Google whether or not the direct partners concerned had considered that the placement clause had prevented them from evaluating the commercial interest in sourcing from its competing intermediaries.

821    First, the replies of [confidential] and [confidential] indicate, in essence, that it was possible to display competing ads, that Google had not requested over several years that the display of those ads be changed and that it was possible to test different locations for Google ads. However, they do not specify whether the location reserved, in practice, for Google ads had prevented direct partners from evaluating the commercial interest in sourcing from its competing intermediaries.

822    Next, the extracts from the replies of the [confidential] group, relied on by Google, are ambiguous. The [confidential] group had indeed answered ‘no’ to the question of whether the placement clause had prevented it from displaying competing ads between 2011 and 2014 or had, in practice, limited its ability to display such ads during the same period. That being said, it should be noted that that group had also wished to qualify that reply, adding the following sentence: ‘However, we refer you to our prior response with respect to such clauses existing in agreements entered into with Google prior to such period.’ Google, however, has not produced that ‘prior response’ to which the said group referred.

823    Last, in respect of Vinden, the statement relied on by Google is limited to describing the content of the placement clause in such an extremely summary manner that it seems impossible to draw consequences from it on the possible foreclosure effects of that clause.

824    That being so, Vinden asserts, before the Court, that the placement clause has not had an effect on its advertising strategy in so far as AFS generated higher revenues than those of other competing online search advertising intermediation services. Nevertheless, it must be held that that assertion does not suffice, even taking into account the replies of [confidential], [confidential], [confidential] and of [confidential], to call into question the tenor of the replies of the [confidential] group and of [confidential], mentioned in recital 499 of the contested decision.

825    In those conditions, as follows from paragraph 802 above, the Commission could rightly take into consideration the examples of replies of direct partners set out in recital 499 of the contested decision as elements that could corroborate its assessment that the placement clause had been capable of preventing direct partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries, even though other direct partners had indicated to it that they had not been affected by that clause.

(2)    Capability of the placement clause to prevent direct partners from being able to adopt certain configurations for their results pages, when the user visited those pages from a desktop device

826    Google claims that most direct partners requested that their websites display more of its ads than the placement clause required. Specifically, it notes that direct partners which on average requested fewer than four of its ads generated less than [confidential]% of AFS revenues between 2011 and 2015. Moreover, the Commission’s file shows that the majority of direct partners requested more than seven Google ads on average. Thus, Google argues that the obligation to display at least three of its ads had had no impact on those direct partners.

827    The Commission disputes Google’s line of argument.

828    In that regard, it should be recalled that, as follows from paragraphs 781 and 782 above, the Commission noted, in recital 500 of the contested decision, that the placement clause had deterred direct partners from sourcing from Google’s competing intermediaries on the ground, inter alia, that that clause prevented those direct partners from being able to adopt certain configurations for their results pages in so far as they by necessity had to display at least three Google ads in a single block when the user visited those websites from a desktop device.

829    According to the Commission, the placement clause deterred in particular direct partners wishing to display only three online search ads or fewer from sourcing from one of Google’s competing intermediaries, since, in that scenario, direct partners could display only Google’s ads.

830    In that regard, it should first be pointed out that the Commission did not identify, in recital 500 of the contested decision, the proportion of direct partners that wished to display only three online search ads or fewer.

831    Second, it must be noted that the circumstance – undisputed by the Commission – that direct partners requesting to display fewer than four online search ads represented less than [confidential]% of AFS’s revenues between 2011 and 2015 is relevant for assessing the foreclosure effect resulting from the requirement to display at least three Google ads. That is all the more the case given that the circumstance – likewise undisputed by the Commission – that the majority of direct partners displayed more than seven Google ads on average suggests that the total number of direct partners wishing to display three online search ads or fewer was low.

832    However, it should be noted that that circumstance concerns only a single aspect of the placement clause, namely the requirement for the direct partner to display at least three Google ads. It thus has no impact on the other aspects of that clause, particularly on the requirement to reserve the most prominent spaces of the results pages of direct partners for Google’s ads and to display them in a block. Consequently, the circumstance noted in paragraph 831 above does not, on its own, mean that, had it not been for that clause, certain direct partners would not have displayed competing ads in spaces generating more revenues and would thus not have sourced a larger part of their requirements from Google’s competing intermediaries.

(3)    Commercial interest in direct partners in sourcing from Google’s competing intermediaries in the absence of the placement clause

833    Google claims that most direct partners would have sourced from it, including in the absence of the placement clause. In that regard, it criticises, in essence, the Commission for not having taken into account the fact that publishers chose AFS, on the ground that it was a service of superior quality, and that publishers therefore had no commercial interest in sourcing from other intermediaries.

834    The Commission disputes Google’s line of argument.

835    First of all, as has been noted in paragraph 108 above, it is apparent from the case-law that, in order to establish that the placement clause was abusive, the Commission was not required to demonstrate that that conduct had actually produced anticompetitive effects, but only that it had been capable of restricting competition in the circumstances of the case.

836    It should be noted that, after having stated, in recital 499 of the contested decision, that the placement clause had prevented direct partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries, the Commission found, in recital 502 of that decision, that at least some direct partners would in any event have had a commercial interest in sourcing from multiple intermediaries absent the placement clause.

837    In that regard, it follows from paragraph 825 above that the Commission was right to consider, in recital 499 of the contested decision, that the fact that the placement clause had prevented direct partners from evaluating the commercial interest in sourcing from other intermediaries allowed it to conclude that that clause could have deterred those direct partners from sourcing from other intermediaries.

838    It follows from paragraphs 825 and 832 above that the evidence produced by Google is not sufficient to find that, had it not been for the placement clause, certain direct partners would not have sourced a larger part of their requirements from Google’s competing intermediaries. The Commission could rightly consider, in recital 500 of the contested decision, that that clause was capable of preventing direct partners from being able to adopt certain configurations for their results pages, when the user visited those pages from a desktop device.

839    In those conditions, it is appropriate to find that the Commission could merely demonstrate that direct partners would have wished to evaluate the commercial interest in sourcing from other intermediaries without having to prove in addition that, had they been able to carry out that evaluation, they would indeed have chosen to source from such intermediaries had it not been for the placement clause.

840    After all, before deciding whether they wanted to source from such an intermediary, direct partners necessarily had to evaluate whether they had a commercial interest in doing so. It follows that, in so far as the placement clause had prevented them from carrying out such an evaluation, the Commission could consider that that clause was capable of restricting competition by deterring them from sourcing from Google’s competing intermediaries.

841    Last, it should be noted that, as is apparent from recital 504 of the contested decision, the fact that Google had included the placement clause in its GSAs was an indication that, notwithstanding the alleged superior quality of AFS, Google considered that, absent that clause, direct partners would have had a commercial interest in sourcing from other intermediaries.

842    Furthermore, [confidential]’s statement appearing in recital 505 of the contested decision, according to which direct partners ‘if possible would like to avoid working with Google’, tends to corroborate the notion that at least some of them had a commercial interest in sourcing from Google’s competing intermediaries.

(4)    Conclusion on the deterrent effect of the placement clause vis-à-vis direct partners

843    It is apparent from the foregoing considerations that, subject to the examination of all the other relevant circumstances, particularly of the duration for which the placement clause applied (see paragraph 848 below), the Commission was right to find that that clause, which prevented direct partners from displaying competing ads on the most prominent spaces of their results pages, could have deterred certain of those direct partners from sourcing at least part of their requirements from Google’s competing intermediaries.

844    First, it should be noted that, as has been mentioned in paragraph 825 above, the Commission could rightly take into consideration the examples of replies of direct partners set out in recital 499 of the contested decision as elements capable of corroborating its assessment that the placement clause could have deterred them from sourcing at least part of their requirements from Google’s competing intermediaries.

845    Second, it must be held that, as has been noted in paragraph 832 above, the mere circumstance that the majority of direct partners – including those generating the most revenues – had decided to display more than three Google ads is insufficient to conclude that, had it not been for that clause, direct partners would not have displayed ads from Google’s competing intermediaries in spaces generating more revenues.

846    Third, it is appropriate to note, as the Commission did in recitals 230 and 276 of the contested decision, that Google’s market share had increased between 2006 and 2016 in the majority of the national markets for online search advertising at issue and in the market for online search advertising intermediation. Thus, in 2016, almost none of Google’s competitors on those markets remained. In addition, the said markets were characterised by the existence of significant barriers to entry and expansion and by a lack of countervailing buyer power on the part of advertisers and publishers. In particular, the effects of scale and network effects had rendered the emergence of new competitors difficult.

847    In those conditions, it is appropriate to find that, contrary to what Google claims, the mere fact that the placement clause had had an effect only on the conduct of certain direct partners does not suffice to demonstrate that that clause had not been capable of restricting competition.

848    Consequently, it is necessary to find that the placement clause could have been capable of producing the foreclosure effect found in the contested decision. However, as has been recalled in paragraphs 776 and 777 above, the question of whether that clause actually had such a capability depends also on the examination of all the other relevant circumstances and, in particular, of the duration for which those direct partners were obliged, in view of the said clause, to reserve the most prominent spaces of their results pages for Google ads, as the Commission found, rightly, in recital 496 of the contested decision.

(b)    Impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

849    In the contested decision, the Commission considered that the placement clause had prevented direct partners from accessing a significant part of the market for online search advertising intermediation. In that regard, first, it noted that the gross revenues generated by GSAs containing that clause represented a significant part of that market. Second, it found that the said clause covered some of the most visited websites in the EEA. Third, it considered that, by requiring direct partners to display at least three ‘wide format’ Google search ads on desktop devices and at least one Google search ad on mobile devices, the clause in question had deprived Google’s competitors of significant revenues derived from the display of such ads. Fourth, it observed that the number of queries carried out on all of the direct partners’ websites constituted a large part of all online search queries carried out in the EEA. Fifth, it took the view that the average duration of GSAs containing the placement clause was long. Sixth, it noted that the fact that that same clause prevented Google’s competitors from accessing a significant part of the said market was consistent with the evolution of Google’s market shares.

(1)    Application of the placement clause to certain online search ad formats

850    Google claims that the placement clause did not apply to certain online search ad formats, such as PLAs, or to online non-search ads. It infers from this that the Commission was wrong to find that that clause reserved the most prominent spaces on direct partners’ results pages for its own online search ads.

851    In that regard, suffice it to hold that, as follows from paragraphs 568 to 571 above, the fact that the placement clause applied neither to certain online search ad formats, including PLAs, nor to online non-search ads does not call into question the fact that that clause prevented Google’s competitors from accessing a significant part of the market for online search advertising intermediation. It should be noted that Google does not prove, including in the context of the first plea, that PLAs, online search ads and online non-search ads belonged to the same market, such that it does not prove that online search advertising intermediation services, online non-search advertising intermediation services and specialised search result advertising intermediation services belonged to the same market, either.

(2)    Market coverage by the placement clause

852    In the contested decision, first, the Commission found that, between 2009 and 2015, on the one hand, the gross revenues generated by GSAs containing the placement clause represented between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA and, on the other hand, those gross revenues and those generated by GSAs containing the exclusivity clause and concluded with all sites direct partners represented, together, between [confidential] and [confidential]% of that market.

853    Second, the Commission noted, referring to the Microsoft study mentioned in paragraph 576 above, that the placement clause covered some of the most visited websites in the EEA. It further specified that the number of online queries carried out on the websites of direct partners constituted a large part of all online search queries made in that area.

(i)    Market share covered by the placement clause

854    First, Google criticises the Commission for having taken into account, for the purposes of assessing the effects of the placement clause, on the one hand, the revenues generated by GSAs containing the exclusivity clause concluded with all sites direct partners and, on the other hand, the revenues generated by GSAs containing the placement clause in which direct partners had not included all of their websites. In that regard, it asserts that the Commission took into account the revenues generated by GSAs concluded with [confidential] and [confidential], when they had never been subject to the placement clause. Second, it claims that the Commission should have taken into account the fact that a significant portion – at least [confidential]% – of the EEA revenues for online search advertising intermediation services remained ‘open’ to its competitors. Third, it notes that the Commission did not assess the coverage of the placement clause in relation to 2016.

855    The Commission contends (i) that Google had progressively started to replace, from 2009, the exclusivity clause with the placement and prior authorisation clauses, (ii) that the gross revenues generated by GSAs containing the placement clause represented a significant part of the market for online search advertising intermediation, and (iii) that it follows from the contested decision that, on 6 September 2016, Google informed the last direct partner of its decision to waive the placement and prior authorisation clauses. It adds that a certain number of direct partners, including significant direct partners, remained party to a GSA containing the placement clause until 3 June 2016 (see paragraph 634 above).

856    In that regard, it is appropriate to recall, as has been noted in paragraph 599 above, that the Commission considered that the exclusivity clause in GSAs concluded with all sites direct partners, on the one hand, and the placement clause, on the other hand, were capable of restricting the part of the market for online search advertising intermediation to which Google’s competitors were able to have access. Moreover, as has been noted in paragraph 601 above, it is settled that those clauses covered different parts of that market simultaneously.

857    Accordingly, as follows from paragraph 603 above, the Commission cannot be considered to have committed an error of law  on the sole ground that it took into account the coverage of the exclusivity clause in GSAs concluded with all sites direct partners in order to determine whether the coverage rate of the placement clause had been sufficient to prevent Google’s competitors from accessing a significant part of the market at issue.

858    It follows that the Commission cannot be criticised for having taken into account GSAs concluded with direct partners, including [confidential] and [confidential], that were not subject to the placement clause, or for having taken into account GSAs containing that clause but in which direct partners had not typically included all of their websites.

859    Furthermore, it has been noted in paragraph 650 above that, even in the scenario most favourable to Google, the combined coverage rate of the exclusivity clause in GSAs in which direct partners had typically included all of their websites, on the one hand, and the placement clause, on the other hand, could be sufficient to be capable of producing a foreclosure effect between 2006 and 2015.

860    In those conditions, it is appropriate to find that the circumstance, cited by Google, that a significant part of the online search advertising intermediation market was not covered by the placement clause is not such as to rule out that that clause was capable of producing a foreclosure effect.

861    However, it is necessary to recall, as has been mentioned in paragraph 641 above, that the Commission has not established that the exclusivity and placement clauses could have prevented Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA in 2016.

(ii) Traffic and number of online queries of the websites covered by the placement clause

862    Google considers, in essence, that, for the reasons set out in paragraph 642 above, the Microsoft study did not enable a causal link to be established between, on the one hand, the fact that the placement clause covered some of the most visited websites in the EEA and, on the other hand, the fact that that clause had prevented its competitors from accessing a significant part of the online search advertising intermediation market.

863    In addition, Google argues that, contrary to what is indicated in the contested decision, the Commission did not demonstrate that the number of online queries made on the websites of direct partners constituted a large part of all online queries made in the EEA. According to Google, Table 27 of that decision is liable to overstate the share of online queries made on the websites of the top 20 direct partners, takes into account websites which were not subject to the placement clause and concerns only the year 2015 and five EEA Member States.

864    The Commission disputes Google’s line of argument.

865    First, in recital 515 of the contested decision, the Commission found, referring to the Microsoft study mentioned in recital 390 of that decision, that ‘some’ of the most visited websites in the EEA were covered by the placement clause.

866    In that regard, it should be noted that, as follows from paragraph 628 above, the Commission determined, in the contested decision, the exact coverage rate of the placement clause on the basis of the gross revenues generated by GSAs containing that clause.

867    As has been noted in paragraph 646 above, the Microsoft study, mentioned in recital 390 of the contested decision, establishes that Google provided online search advertising intermediation services to between [confidential] and [confidential]% of the most visited web domains in Germany, Spain, France, Italy and the United Kingdom in 2010. It is true that, as Google notes, that study concerns only one year of the period of infringement and five Member States. Likewise, Google is justified in noting that that study does not enable the traffic generated specifically by the websites covered by the placement clause to be identified. Last, it rightly observes that the number of website visits is not necessarily a reliable indication of online search advertising revenues.

868    However, on the one hand, the fact remains that the Microsoft study constitutes an additional indication enabling an assessment of the scale of the online search advertising intermediation services provided by Google and the coverage of the market for online search advertising intermediation by that clause, in so far as, first of all, it involves five of the largest Member States of the EEA, next, it is not disputed that at least some of the websites covered by that study were subject to the placement clause and, last, there is some correlation between the number of website visits and the revenues from online search advertising.

869    On the other hand and in any event, the Commission did not base its calculation of the coverage rate of the placement clause on the Microsoft study, such that, even assuming that Google’s arguments in that regard were founded, that would have no effect on the calculation of that rate performed by the Commission.

870    Second, it is appropriate to note that, in Table 27 of the contested decision, the Commission excluded the online queries made on the search engines of Google, Microsoft, Yahoo!, Yandex and Baidu as well as the online queries made on sites belonging to Microsoft and to Yahoo! from the total number of online queries made in the EEA. Thus, contrary to what Google suggests, the Commission could not exclude those same online queries from the total number of online queries made on the websites of the top 20 direct partners, since it is undisputed that no GSA applied to the websites on which those online queries were made.

871    However, it should be noted that Google argues, rightly, that Table 27 of the contested decision concerned only one year of the period of infringement between 31 March 2009 and 6 September 2016 and that it concerned only 5 of the 31 Member States of the EEA during that period. Moreover, it should be borne in mind that the Commission acknowledges that 3 of the 20 direct partners that it identified were not subject to the placement clause. On the one hand, however, it did not specify the proportion of online queries generated by the websites of those three direct partners. On the other hand, while it argues that two of those three direct partners were subject to the exclusivity clause, it is worth noting that that circumstance does not figure among the grounds of the contested decision and that, in any event, the Commission does not claim that those two direct partners figured among the all sites direct partners identified in recital 348 of that decision.

872    Accordingly, it is appropriate to find that the circumstance, mentioned in recital 518 of the contested decision, according to which the number of online queries made on the websites of direct partners constituted a large part of all online queries made in 2015 in Germany, Spain, France, Italy and the United Kingdom does not suffice to demonstrate that the number of online queries made on websites specifically subject to the placement clause constituted a large part of all online queries made between 31 March 2009 and 6 September 2016 in the EEA.

873    Nevertheless, it should be noted that the question of whether the websites covered specifically by the placement clause had generated a high number of online queries serves as merely one indicator enabling assessment of the extent of the coverage of the market for online search advertising intermediation as a whole by that clause.

874    As has been recalled in paragraph 866 above, the Commission moreover determined the coverage rate of the placement clause on the basis of the gross revenues generated by GSAs containing that clause. Furthermore, it has been noted in paragraphs 650 and 859 above that that coverage rate, combined with that of the exclusivity clause in GSAs in which direct partners had typically included all of their websites, could be sufficient for those clauses to be capable, together, of producing a foreclosure effect.

875    Accordingly, it is appropriate to find that the Commission could rely on the Microsoft study and on the data recorded in Table 27 of the contested decision as indications corroborating the assessment of the coverage rate of the placement clause, without it being necessary to rule on the admissibility of Google’s line of argument, disputed by the Commission.

(3)    As-efficient competitor test

876    In the contested decision, the Commission found that the placement clause was capable of foreclosing a hypothetical competing intermediary as efficient as Google. First, the revenues generated between 2009 and 2015 by GSAs containing that clause represented between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA. Second, the revenues generated by that clause and the exclusivity clause in GSAs concluded with all sites direct partners between 2009 and 2015 represented [confidential] to [confidential]% of that market. Third, Google held a ‘very large’ share of the market between 2006 and 2016. Fourth, that market was prone to network effects.

877    The Commission moreover indicated that it was ‘doubtful’ whether a hypothetical intermediary as efficient as Google could have emerged during the period of application of the placement clause. It last found that the question of whether Google had pursued a strategy aimed at excluding competitors as efficient as it was irrelevant.

878    Google argues that the Commission failed to prove that a competitor as efficient as it could not have emerged on the market for online search advertising intermediation owing to the placement clause. Vinden states in that regard that that clause did not prevent it from sourcing from Yahoo!. Last, Google criticises the Commission for having considered that the absence of a strategy aimed at excluding competitors as efficient as it was irrelevant in this case.

879    The Commission disputes Google and Vinden’s line of argument.

880    In that regard, it should be noted that, as has been mentioned in paragraphs 656 and 657 above, the fact that the Commission indicated that it was ‘doubtful’ whether a hypothetical intermediary as efficient as Google could have emerged during the period of application of the placement clause is not such as to call into question the legality of the contested decision.

881    Moreover, as has been noted in paragraph 665 above, the Commission could merely demonstrate the capability of the exclusivity clause in GSAs concluded with all sites direct partners to produce a foreclosure effect by relying on several relevant elements, without necessarily relying, to that end, on the as-efficient competitor test. Furthermore, as has been noted in paragraph 671 above, it is settled that Google did not provide during the administrative procedure – or before the Court – any analysis based on that test.

882    In the case at hand, the very substance of the placement clause, which prevented, in principle, direct partners from displaying competing ads in the most prominent spaces of their results pages, and the factors mentioned in recital 549 of the contested decision, namely, on the one hand, the fact that that clause, together with the exclusivity clause in GSAs in which direct partners had typically included all of their websites, covered a significant part of the online search advertising intermediation market, as has been essentially noted in paragraphs 650 and 859 above, and, on the other hand, the extent of Google’s dominant position resulting in particular from its very high market shares and from the barriers to entry and expansion in the form inter alia of network effects, could demonstrate that the placement clause could be capable of foreclosing a hypothetical competitor as efficient as Google. It follows that those factors could also demonstrate that it was ‘doubtful’ that such a competitor could have emerged during the period of application of the said clause.

883    Last, as follows from paragraphs 678 to 681 above, the Commission cannot be criticised, on the one hand, for not having established that Google had adopted a strategy aimed at excluding competitors at least as efficient as it and, on the other hand, for not having taken into account the fact that it did not intend to exclude such competitors.

(4)    Duration of GSAs and the unilateral termination right of some direct partners

884    In the contested decision, the Commission considered that the average duration of GSAs containing the placement clause was long. It stated, in that regard, that Google and direct partners had extended certain GSAs, sometimes several times, without substantial modifications. In addition, it noted that the period during which the direct partners were required to source a minimum number of Google ads and at the same time reserve the most prominent spaces for Google ads was also long. Last, it found that only one direct partner held a unilateral termination right.

885    Google criticises the Commission for having confused the length of the commercial relationship with the direct partners subject to the placement clause with the duration of the GSAs containing that clause. It thus notes that the GSAs mentioned in footnotes 707 and 713 to the contested decision had a duration of two years or less between each renewal or each extension. In addition, it states that the Commission failed to take into account the fact that some direct partners held a unilateral termination right.

886    Vinden maintains that the GSAs concluded with Google were short, being renewed every two years on average. Additionally, it notes that the order form for 2011, which it had completed, allowed for a unilateral termination right if notice was given 60 days prior to the one-year anniversary of the contract. Thus, it claims that it had had frequent opportunities to decide to source from one of Google’s competing intermediaries.

887    First, the Commission contends that the GSAs of 11 of the 15 direct partners mentioned in footnotes 707, 713, 766 and 767 to the contested decision, including those concluded with Vinden, were always extended before they ‘came up for renewal’. Google’s competitors therefore were ‘never able to contest that demand’. Moreover, the Commission notes that most of the GSAs concluded by one of the remaining four direct partners were extended before that point, such that the said competitors of Google were ‘mostly not able to contest that demand’.

888    Second, first of all, the Commission considers that the fact that some direct partners held a unilateral termination right was irrelevant given that the existence of that right did not prevent the application of the placement clause until those direct partners exercised that right, where applicable. Next, it states that the unilateral termination rights at issue could not be exercised at any moment. Last, it notes that the argument alleging that the six direct partners identified by Google in the reply had such a right is belated and, therefore, inadmissible.

889    First of all, regarding the inadmissibility argument raised by the Commission, it is appropriate to note that Google criticises, in paragraph 131 of the application, the contested decision for not having taken into account the fact that some direct partners had a unilateral termination right. It refers, in that regard, to Tables 1 and 2 of Annex A.55 to the application, which identify the direct partners, the GSAs of which are mentioned in footnotes 707, 713, 766 and 767 to the contested decision, enjoying such a right. In those conditions, having regard to the case-law recalled in paragraph 691 above, it should be noted that Google’s argument, raised in the reply and alleging that other direct partners also enjoyed that right, presents a sufficiently close connection with the argument developed in the application in order to be considered as forming part of the normal evolution of debate in proceedings before the Court.

890    As for the merits of Google’s line of argument, it is appropriate to note that, as has been indicated in paragraph 778 above, the Commission was right to find that the placement clause was akin to a relaxed exclusivity clause so far as concerns websites that were included in GSAs containing that clause since it reserved the most prominent spaces of direct partners’ results pages for Google ads.

891    In those conditions, it must be held that, as follows from paragraphs 695 and 696 above, the duration of the obligation by which direct partners had to reserve the most prominent spaces of their results pages counts among the circumstances relevant for assessing the foreclosure effect of that clause.

892    In that regard, first, it should be noted that, in order to consider that the average duration of GSAs containing the placement clause was long, the Commission relied only, in recital 519 of the contested decision, on the fact that many GSAs had been extended, sometimes several times, without substantial modifications.

893    On the one hand, it is apparent from Annex A.55 to the application, the content of which is not disputed by the Commission, and from recitals 519 and 525 of the contested decision, and more specifically in footnotes 707 and 713 to that decision, the content of that latter footnote being restated in footnotes 766 and 767 to the said decision mentioned in the Commission’s written submissions, that that institution took into account the total duration of those GSAs, by including all of their extensions, where applicable. However, it took into account neither the initial duration of each of the said GSAs, taken individually, nor the duration of each of their extensions, where applicable, which paragraph 241 of the defence essentially confirms.

894    On the other hand, it is apparent from footnote 707 to the contested decision and from Annex A.55 to the application that, contrary to what recital 519 of that decision suggests, the Commission did not only take into account the total duration of each GSA, including extensions, but also the cumulative duration of the different GSAs concluded by the same direct partner.

895    Thus, it is apparent from the contested decision and from Annex A.55 to the application that the Commission took into account neither the duration of each of the GSAs, taken individually, nor the duration of each of the possible extensions of those GSAs.

896    Second, to justify the lack of taking into account of unilateral termination rights for the purposes of its assessment, the Commission found, in recital 526 of the contested decision, that Google had identified, during the administrative procedure, only one direct partner which enjoyed such a right. It thus considered it an exception, asserting that no other direct partner had such a right.

897    However, it is apparent from Annexes A.55 to the application and C.8 to the reply that at least nine other direct partners, namely Vinden, [confidential], [confidential], [confidential], [confidential], [confidential], [confidential], [confidential] and [confidential], held a unilateral termination right, which the Commission does not dispute. Moreover, it is apparent from footnotes 707, 713, 766 and 767 to the contested decision that the Commission relied on the duration of GSAs concluded with at most 15 direct partners. Accordingly, the fact that at least 10 direct partners enjoyed a unilateral termination right cannot be regarded as an exception justifying that it should not be taken into account for the purposes of the assessment of the foreclosure effect of the placement clause.

898    In those conditions, the Commission could not, solely on the basis of the considerations recalled in paragraphs 892 and 896 above and without having examined the actual conditions and the terms under which the extensions of the GSAs had been agreed, as well as the substance of the clauses providing for the unilateral termination rights held by some of the direct partners and the conditions in which those rights could be exercised, exclude that those direct partners had the option of sourcing from Google’s competing intermediaries in respect of the most prominent spaces of their results pages, including before any extension of their GSAs, or before a unilateral termination right had been exercised. It follows that the Commission also could not find that those intermediaries had not had the possibility of disputing the part of the market for online search advertising intermediation covered by GSAs containing the placement clause for the total duration of those GSAs, let alone for their cumulative duration.

899    That conclusion is not called into question by the Commission’s arguments.

900    In the first place, the Commission contends that certain GSAs were extended before they reached their term. However, it should be recalled that, as has been mentioned in paragraph 707 above, the Commission relies on the unsubstantiated premiss according to which the other intermediaries could not compete with Google when those GSAs were being renewed. In particular, it is appropriate to point out that the Commission puts forward no evidence proving that the negotiations on the extension of a GSA could not be done at the end of a competitive process by which the direct partner concerned compared the services provided by Google and by its competitors.

901    In the second place, it should be noted that the Commission is not justified in arguing, in the defence, that it was irrelevant whether direct partners had unilateral termination rights, on the ground that the placement clause continued to apply until that right was exercised. As has been noted in paragraph 714 above, the Commission had to examine the substance of the clauses providing for those rights, the conditions in which they could be exercised as well as which direct partners could enjoy them in order to determine whether the said rights were capable of calling into question, at least to a certain extent, the fact, found in the contested decision, that the placement clause had prevented Google’s competitors from accessing a significant part of the market for online search advertising intermediation for the duration of GSAs containing that clause.

(5)    Conclusion on the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

902    As has been found in paragraph 859 above, the Commission was right to consider that, having regard to the coverage of the exclusivity clause in GSAs concluded by all sites direct partners, in the light of the circumstances recalled in paragraph 602 above, the coverage of the placement clause could be sufficient to enable that clause to be capable of producing a foreclosure effect between 31 March 2009 and 31 December 2015. In addition, it has been noted in paragraph 882 above that Google was not justified in arguing that the Commission had failed to prove that a competitor as efficient as it could not emerge on the market for online search advertising intermediation or that such a competitor was capable of being excluded from that market.

903    However, on the one hand, as follows from paragraph 861 above, the Commission did not establish that the placement clause could have produced a foreclosure effect, owing to its coverage, between 1 January and 6 September 2016. On the other hand, as follows from paragraph 898 above, it failed to take into consideration all the relevant circumstances of the case in the context of the assessment of the duration for which direct partners were obliged, by virtue of that clause, to reserve the most prominent spaces of their results pages for Google ads.

904    It follows that, contrary to what the case-law recalled in paragraph 107 above requires, the Commission has not demonstrated to the requisite legal standard, in the light of all the circumstances of the case, that the placement clause was capable of preventing Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA, for the duration for which that clause applied.

(c)    Conclusion on the second part of the third plea

905    As has been recalled in paragraph 769 above, in recital 494 of the contested decision, the Commission considered that the placement clause was capable of restricting competition, in the light of all the relevant circumstances of the case. It noted in that regard that that clause had (i) deterred direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers.

906    As has been recalled in paragraphs 770 and 771 above, the Commission essentially considered that the placement clause was capable of producing a foreclosure effect, finding that it had, on the one hand, deterred direct partners from sourcing from Google’s competing intermediaries and had, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation. In addition, it inferred from that foreclosure effect that that clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

907    As has been recalled in paragraph 776 above, in recital 496 of the contested decision, before examining the effects of each of the five restrictions of competition that it identified (see paragraph 905 above), the Commission specified that, for the purposes of its analysis seeking to demonstrate that the placement clause was capable of restricting competition, it had taken into account the ‘duration of [that c]lause’, thus highlighting, essentially, the importance of the developments set out in recitals 519, 525 and 526 appearing in Section 8.4.4.2 of that decision, relating to the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation. It also indicated that it had taken into consideration the coverage rate of the said clause, which it examined in the same section of the said decision. It is also apparent from the systemic place of the said recital 496 in the structure of the contested decision that the Commission took into account that duration and that coverage rate when it examined the effects of the placement clause in the context of each of the five restrictions identified in the contested decision.

908    As has been found in paragraph 903 above, however, the Commission failed to take into consideration all the relevant circumstances of the case in the context of the assessment of the duration for which direct partners had been obliged to reserve the most prominent spaces of their results pages for Google ads by virtue of the placement clause.

909    Moreover, as has also been found in paragraph 903 above, the Commission did not establish that the placement clause could have produced a foreclosure effect, owing to its coverage, between 1 January and 6 September 2016.

910    It follows that the errors committed by the Commission, recalled in paragraphs 908 and 909 above, vitiate all of the restrictions identified by it in the contested decision, such that it is appropriate to conclude that the Commission has not demonstrated, to the requisite legal standard, that the placement clause had been capable of deterring direct partners from sourcing from Google’s competing intermediaries or that it had been capable of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA and, consequently, that that clause had been capable of having the foreclosure effect found in that decision.

911    In those conditions, it is appropriate to find that the Commission has also not demonstrated, to the requisite legal standard, that the placement clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

912    Moreover, it must be stated that, in recitals 541 and 542 of the contested decision, the Commission essentially found that the binding nature of the mock-ups had exacerbated the foreclosure effect of the placement clause by further limiting the possibility for direct partners to modify the positioning of both Google ads and competing ads. However, the Commission has neither established that the placement clause was capable of having such an effect nor alleged that the mock-ups alone were capable of having that effect. Accordingly, the mock-ups cannot suffice, on their own, to demonstrate that the placement clause constituted an infringement of Article 102 TFEU.

913    It follows that the second part of the third plea must be upheld, without it being necessary to rule on the other arguments of Google under that plea, and, consequently, the contested decision must be annulled to the extent that it found that the placement clause constituted an infringement of Article 102 TFEU.

E.      Fourth plea: the prior authorisation clause did not constitute an abuse of a dominant position

914    By the fourth plea, Google criticises the Commission for having considered that the prior authorisation clause constituted an abuse of a dominant position within the meaning of Article 102 TFEU. This plea consists of two parts, the first alleging that there was no restriction of competition, and the second alleging that the said clause was objectively justified.

915    As a preliminary point, it must be recalled that, in the March 2009 template GSA, the prior authorisation clause was worded as follows:

‘Unless approved in writing in advance by Google, Company will not make any changes in relation to: … the display of Equivalent Ads, AFS Ad Sets or AFS Ads on a Results Page, including changes to their number, colour, font, size or placement or the extent to which they are clickable.’

916    The prior authorisation clause was supplemented by clause 6.2(b) of the March 2009 template GSA which was worded as follows:

‘Where Company requests approval pursuant to [the prior authorisation clause] above, Google may only withhold its approval on grounds that the proposed change would be in breach of the applicable Agreement or the Google Branding Guidelines and Google may not withhold its approval on purely commercial grounds. If Google does not respond to any such request for approval within 15 business days of receipt from Company, such approval shall be deemed given by Google.’

917    As has been recalled in paragraph 730 above, clause 1.1 of the March 2009 template GSA stipulated that the expression ‘Equivalent Ad’ had to be understood as referring to ‘any advertisements that are the same as or substantially similar in nature to the AFS Ads provided by Google under any Agreement’.

918    The wording of the prior authorisation clause was modified in the template GSA over time. That clause was also worded in the following manner:

‘If Company wishes to make changes in relation to the display of: Equivalent Ads on a Results Page, including changes to their number, colour, font, size or placement or the extent to which they are clickable, Company will not make any changes unless approved in writing in advance by Google. Google may not withhold its approval unless such proposed change would be in breach of the applicable Agreement. If Google does not respond to any request for approval set out in this clause … within 15 business days of receipt from Company, such approval shall be deemed given by Google.’

919    In the contested decision, the Commission considered, in recital 573 thereof, that the prior authorisation clause was capable of restricting competition, in the light of all the circumstances of the case. In that regard, it noted that that clause had (i) deterred direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers.

920    More specifically, it should be noted that, in finding that the prior authorisation clause had, on the one hand, deterred direct partners from sourcing from Google’s competing intermediaries and, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, the Commission essentially considered that that clause was capable of producing a foreclosure effect.

921    In addition, it should be noted that the Commission inferred from the foreclosure effect of the prior authorisation clause that that clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

922    First, it follows from recitals 598 to 600 of the contested decision that the foreclosure effect of the prior authorisation clause had deterred Google’s competing intermediaries from providing or developing different online search ads, with the result that that clause had deterred them from investing in innovation. Next, it follows from recital 602 of that decision that that effect had deprived the said intermediaries of revenues and data that they could have used to provide online search ads. Last, it follows from recital 605 of the same decision that the said effect had allowed Google to set the prices paid by advertisers at a high level, thereby increasing the prices consumers paid for the goods featured in the online search ads. The Commission added, in recital 606 of the contested decision, that the fact that the prior authorisation clause had possibly deterred innovation had also deprived consumers of a wider choice of online search ads.

923    In the context of the first part of the fourth plea, Google submits that the prior authorisation clause, first, did not produce the foreclosure effects found in the contested decision; second, had not helped it to maintain or strengthen its dominant position on the national markets for online search advertising at issue; and, third, had neither deterred innovation nor harmed consumers.

924    Surfboard and Vinden claim that the prior authorisation clause had no effect on their conduct and that it was objectively justified.

925    It is appropriate at the outset to examine the foreclosure effect identified in the contested decision resulting from the prior authorisation clause. It must therefore be ascertained whether that clause was capable, on the one hand, of deterring direct partners from sourcing from Google’s competing intermediaries and, on the other hand, of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation.

926    In that regard, it should be noted that, in recital 574 of the contested decision, the Commission specified that, for the purposes of its analysis seeking to demonstrate that the prior authorisation clause was capable of restricting competition, it had taken into account all the relevant circumstances, including, on the one hand, the extent of Google’s dominant position, both on the national markets for online search advertising at issue and on the market for online search advertising intermediation, and, on the other hand, the share of the latter market covered by the said clause and the ‘duration of [that c]lause’. It referred, in that regard, respectively, to Section 7 of that decision, relating to Google’s dominant position, which includes the considerations recalled in paragraphs 401 to 404 above, and to the whole of Section 8.5.4.2 of the said decision, relating to the impossibility for Google’s competing intermediaries to access a significant part of the said market.

927    In line with what has been found in paragraphs 400 and 405 above, it should be found, on the one hand, that the approach followed by the Commission is in conformity with the case-law and, on the other hand, that Google does not challenge the content of Section 7 of the contested decision other than by arguing, under the first plea, that the Commission erroneously defined the relevant markets in Section 6 of that decision.

928    Taking into account those factors, it is appropriate to examine, first, whether the prior authorisation clause was capable of deterring them from sourcing from Google’s competing intermediaries and, next, whether the said clause was capable of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation.

1.      Deterrent effect of the prior authorisation clause vis-à-vis direct partners

929    In the contested decision, the Commission found that the prior authorisation clause had deterred direct partners from sourcing from Google’s competing intermediaries.

930    Specifically, the Commission found (i) in recital 577 of the contested decision, that the prior authorisation clause had prevented direct partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries, by imposing a ‘triangular’ negotiation in order to make changes to the display of competing ads, (ii) in recital 578 of that decision, that the scope of that clause and Google’s refusal to discuss or clarify it had deterred direct partners from sourcing from those intermediaries, and (iii) in recital 579 of the said decision, that, absent that same clause, direct partners would have sourced from the said intermediaries more freely.

931    Google disputes the content of each of recitals 577, 578 and 579 of the contested decision.

932    In the first place, Google criticises the Commission for having found, in recital 577 of the contested decision, that the prior authorisation clause had prevented direct partners from evaluating the commercial interest in sourcing from other intermediaries.

933    More specifically, first, Google claims that the prior authorisation clause, on the one hand, applied only to the websites that direct partners chose to include in their GSAs and, on the other hand, allowed the display of competing ads. Moreover, the Commission identified no instance of Google refusing the display of such ads. Second, it notes that the said clause applied neither to the initial display of competing ads nor to the compliance of that display with its various ad policies (and seeking to prohibit the display of offensive, unsafe, undesirable or deceptive ads), such that that same clause could not have deterred direct partners from displaying competing ads. Third, it states that, while the clause in question had been intended to prevent the changes proposed by direct partners from breaching the aforementioned policies, the Commission did not consider those same policies to be capable of restricting competition. Fourth, it criticises the Commission for not having taken into account, on the one hand, the fact that, by virtue of clause 6.2(b) of the March 2009 template GSA, it could not reject a request for a change relating to the display of ads on purely commercial grounds and, on the other hand, evidence showing that it had accepted requests from direct partners to change that display. In that regard, it notes moreover that the Commission cited no instance of it refusing to change the display of competing ads by virtue of that clause. Fifth, it claims that the Commission also ignored evidence showing that direct partners could evaluate competing ads shown on websites subject to the prior authorisation clause.

934    In the second place, Google criticises the Commission for not having taken into account, in recital 578 of the contested decision, evidence demonstrating that direct partners could negotiate the removal of the prior authorisation clause from their GSAs. In addition, it states that no direct partner statement, quoted by the Commission in that recital, can establish that it refused to discuss or clarify the scope of that clause.

935    In the third place, Google criticises the Commission for having relied, in recital 579 of the contested decision, on statements from direct partners which could not confirm that they would have sourced from other intermediaries more freely had it not been for the prior authorisation clause.

936    Surfboard maintains that the prior authorisation clause promoted a positive user, advertiser and direct partner experience, while preventing the display of offensive and deceptive ads. It also notes that Google could not refuse a change request on purely commercial grounds. Moreover, it submits that that clause did not prevent it from making the changes that it wanted on its websites, as it had explained to the Commission during the administrative procedure. Vinden adds that the said clause had not required it to seek Google’s permission in order to change the layout of competing ads.

937    The Commission disputes Google, Surfboard and Vinden’s line of argument.

938    As a preliminary point, it should be borne in mind that the prior authorisation clause, associated with clause 6.2(b) of the March 2009 template GSA, provided that direct partners had to obtain prior written approval from Google or, where it did not respond to their request for approval, await the expiry of a period of 15 business days, before changing the display of competing ads, including in terms of number, colour, font, size, placement and the extent to which they were clickable.

939    Thus, the prior authorisation clause enabled Google, on the one hand, to be informed of any intention to change how competing ads were displayed and, on the other hand, to monitor continuously direct partners’ commercial relations with competing intermediaries and interfere with them, by having the option, where appropriate, not to authorise, under the conditions prescribed by clause 6.2(b) of the March 2009 template GSA, changes to the display of those ads which were not to its liking.

940    The Commission was therefore right to note, in recital 577 of the contested decision, that, by requiring direct partners to seek prior approval from Google before making any change to the display of competing ads, the prior authorisation clause imposed a more burdensome triangular negotiation between those direct partners, Google and its competing intermediaries.

941    Similarly, it should be noted that the Commission considered, rightly, in recital 579 of the contested decision, that, absent the prior authorisation clause, direct partners could have sourced from other intermediaries more freely, notwithstanding Google’s line of argument seeking to call into question the relevance of the statements of direct partners quoted, by way of example, in that recital.

942    Furthermore, it is appropriate to note that the reply of [confidential], quoted in recital 578 of the contested decision, is such as to confirm the fact that Google had refused to negotiate the scope of the prior authorisation clause. It is apparent from that recital – and undisputed by Google – that [confidential] had proposed amending its GSA to specify, first, that Google would act ‘reasonably’ and, second, that it could refuse a request made under that clause only by reference to the ‘Google brand guidelines’ and ‘other Google policies’. It follows that [confidential] attempted to amend the clause in question in order to limit the circumstances in which Google could refuse changes to the display of competing ads. Google did not accede to that request, however.

943    What is more, as is apparent from footnote 422 to and from recital 630 of the contested decision and as Google confirms in its written submissions, all GSAs that included the prior authorisation clause also contained the placement clause. It has been noted in paragraph 843 above, however, that the placement clause was capable of deterring certain direct partners from sourcing at least part of their requirements from Google’s competing intermediaries. In those conditions, the prior authorisation clause should be regarded as having been able to exacerbate the deterrent effect of the placement clause by making it more difficult to change the display of competing ads already restricted by that clause.

944    Accordingly, it is appropriate to find that, in the light of the foregoing and of the extent of Google’s dominant position on the market for online search advertising intermediation, and particularly of the existence of significant barriers to entry and expansion on that market, the Commission has demonstrated to the requisite legal standard that the prior authorisation clause, in so far as it obliged direct partners to obtain prior written approval from Google before changing the display of competing ads, which forced them to engage in triangular negotiations with interference by Google and prevented them from sourcing from its competitors more freely, was able to deter certain direct partners from sourcing at least part of their requirements from Google’s competing intermediaries.

945    Google’s other arguments are incapable of calling that assessment into question.

946    First, it should be noted, like the Commission, that, as follows inter alia from paragraph 787 above, the fact that the prior authorisation clause applied only to websites included in GSAs containing that clause and not to all of the websites of direct partners is irrelevant to determining whether the said clause had prevented those partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries, at least as regards websites included in those GSAs.

947    Second, it should be noted that the circumstance, cited by Google, that the prior authorisation clause was, in essence, objectively justified, on the ground that it sought to prohibit the display of offensive, unsafe, undesirable or deceptive ads, related not to the existence of anticompetitive effects as such. It is only in the event that it is found, after having assessed all the relevant circumstances, that that clause produced such effects that it will be appropriate to ascertain whether Google has justified the use of the said clause (see, to that effect, judgment of 27 March 2012, Post Danmark, C‑209/10, EU:C:2012:172, paragraph 40).

948    Third, it should be noted that Google is not justified in arguing that clause 6.2(b) of the March 2009 template GSA was liable to call into question the deterrent effect of the prior authorisation clause.

949    It is true that clause 6.2(b) of the March 2009 template GSA provided a framework for the possibility for Google to refuse a change request relating to the display of competing ads submitted by a direct partner. It also stipulated that such a request was deemed to have been approved if Google did not respond to it within a period of 15 business days. However, its mere presence in GSAs does not call into question the fact that the prior authorisation clause required direct partners to request prior approval from Google in order to change the display of competing ads and that it thus made it more difficult in practice for those direct partners to source from other intermediaries. Such was the case, irrespective of the question of knowing under what circumstances Google could refuse requests to change the display of competing ads and of the possibility for it to delay the implementation of such changes having regard to the period of 15 business days prescribed for responding to such requests. Moreover, it is true, as Google claims, that the prior authorisation clause and clause 6.2(b) of the March 2009 template GSA did not relate to the initial display of competing ads. However, it is appropriate to state that that clause made it more difficult to change the display of competing ads, once those ads had been displayed on a results page of a direct partner’s website.

950    Fourth, there are certainly grounds for noting, as Google does, that the Commission did not identify any cases in which Google denied, on the basis of the prior authorisation clause, a change request submitted by a direct partner. Likewise, Google is right to maintain that the Commission’s file contained evidence that it had in fact acceded to such requests. However, the Commission did not contend, in the contested decision, that the restriction of competition that it had found stemmed from Google’s refusal to authorise such requests. On the contrary, it considered, rightly, that the mere act of placing direct partners under the obligation to obtain prior written approval from Google or, failing that, to wait 15 business days before being able to change the said display resulted in their being forced to participate in a more burdensome triangular negotiation to that end and prevented them from sourcing from Google’s competitors more freely.

951    Fifth, Google cites inter alia the replies to a request for information of 30 October 2015 of three direct partners, namely [confidential], [confidential] and [confidential], which informed the Commission that they had been able to assess the performance of the competing ads displayed on websites of theirs that were subject to the prior authorisation clause. However, on the one hand, it should be noted that, in its written submissions, Google does not specify whether [confidential], [confidential] and [confidential] had to obtain its approval before making such assessments. Furthermore, it is apparent from those written submissions that the scope of those assessments was limited, [confidential] having stated that it had carried out ‘only … one small test’ with another provider and [confidential] having specified that, after ‘step-by-step negotiation over years’, it had been allowed ‘to use a percentage of [its] users in so called “test buckets” to have a certain amount of traffic on the alternative to be tested’.

952    On the other hand, it is not disputed that, by virtue of that clause, direct partners had to obtain Google’s approval before changing the display of competing ads. It follows that, even assuming that those three direct partners had been able to assess the performance of competing ads, that does not permit the conclusion that the said clause had not had a deterrent effect on direct partners.

953    Sixth, while Google is right to note that certain direct partners had been able to negotiate the scope – or even the removal – of the prior authorisation clause, it follows from paragraph 942 above that other direct partners had not had that possibility. In addition, the fact that certain direct partners were not subject to that clause does not, in any event, rule out that that same clause had been able to deter the direct partners that were subject to it from sourcing a larger part of their requirements from Google’s competing intermediaries. In those conditions, it is necessary to find that the said clause could have been capable of producing the foreclosure effect found in the contested decision. However, as has been recalled in paragraphs 926 and 927 above, the question of whether that clause actually had such a capability depends also on the examination of all the other relevant circumstances and, in particular, of the duration for which those direct partners were obliged, in view of the said clause, to request authorisation from Google to change the display of competing ads, as the Commission found, rightly, in recital 574 of the contested decision.

2.      Impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

954    In the contested decision, the Commission considered that the prior authorisation clause had prevented direct partners from accessing a significant part of the market for online search advertising intermediation. In that regard, first, it noted that that clause afforded Google the right to control changes to the display of online search ads that had an impact on CTR, namely changes to those ads’ number, colour, font, size or placement, as well as changes to ‘the extent to which [search ads] were clickable’. Second, it found that Google had gradually included the said clause in the overwhelming majority of GSAs. Third, it considered that the gross revenues generated by GSAs containing that clause represented a significant part of that market. Fourth, it noted that the said clause covered some of the most visited websites in the EEA. Fifth, it observed that the number of queries carried out on direct partners’ websites constituted a large part of all online search queries carried out in the EEA. Sixth, it took the view that the average duration of GSAs containing the clause in question was long. Seventh, it indicated that the fact that that same clause prevented Google’s competitors from accessing a significant part of the said market was consistent with the evolution of Google’s market shares.

(a)    Market coverage by the prior authorisation clause

955    In the contested decision, first, the Commission considered that, between 2011 and 2015, the gross revenues generated by GSAs containing the prior authorisation clause represented, on the one hand, between [confidential] and [confidential]% of the gross revenues generated by all GSAs in the EEA and, on the other hand, between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA.

956    Second, the Commission noted, referring to the Microsoft study mentioned in paragraph 576 above, that the prior authorisation clause covered some of the most visited websites in the EEA. In addition, it considered, on the basis of Table 27 of the contested decision, mentioned in paragraphs 870 and 871 above, that the number of queries carried out on direct partners’ websites constituted a large part of all online search queries carried out in the EEA.

(1)    Market share covered by the prior authorisation clause

957    Google argues that the Commission failed to establish that the prior authorisation clause covered a significant part of the market for online search advertising intermediation in the EEA.

958    First, it is apparent from the Commission’s own calculations that the prior authorisation clause covered only [confidential]% and [confidential]% of that market in 2009 and in 2010, respectively. In addition, the gross revenues generated by GSAs containing that clause represented, on average, between 2009 and 2015, only [confidential]% of the said market and, at most, less than [confidential]% of it. Last, Google notes that the Commission did not assess the coverage of the said clause in relation to 2016.

959    Second, Google criticises the Commission for having combined, for the purposes of the assessment of the effects of the prior authorisation clause, the coverage rate of that clause with that of the placement clause and the exclusivity clause in GSAs concluded with all sites direct partners. In that regard, it notes that the Commission took into account revenues from direct partners which were never subject to the prior authorisation clause and which had typically not included all of their websites in their GSAs.

960    On the one hand, the Commission contends that Google had progressively started to replace, from 2009, the exclusivity clause with the placement and prior authorisation clauses and that, consequently, the fact, cited by Google, that the gross revenues generated by GSAs containing the prior authorisation clause represented, on average, during the period of infringement, [confidential]% of the market for online search advertising intermediation was ‘misleading’. It recalls, in that regard, that the gross revenues generated by GSAs containing the placement clause represented between [confidential] and [confidential]% of that market and that that clause covered some of the most visited websites in the EEA. On the other hand, it notes that it follows from the contested decision that, on 6 September 2016, Google had informed the last direct partner of its decision to waive the placement and prior authorisation clauses. It adds that a certain number of direct partners, including significant direct partners, were party to a GSA containing the prior authorisation clause until 3 June 2016.

961    In that regard, first, it should be recalled that the Commission found, in the contested decision, that the prior authorisation clause had constituted an abuse of a dominant position from 31 March 2009 to 6 September 2016.

962    Second, it should be noted that, in the contested decision, the Commission compared, in recitals 586 and 587 thereof, the proportion of revenues generated by GSAs containing the prior authorisation clause between 2009 and 2015, on the one hand, with the revenues generated by all GSAs and, on the other hand, with the revenues generated in the market for online search advertising intermediation in the EEA. From this it inferred that the rate of coverage of that market by that clause had been ‘significant’ between 2011 and 2015.

963    In addition, it follows from recital 611 of the contested decision, concerning the as-efficient competitor test, that the Commission found that the prior authorisation clause had been capable of producing a foreclosure effect in respect of such a competitor between 2009 and 2015, regard being had to the combined coverage rate of that clause, to the exclusivity clause in GSAs concluded with all sites direct partners, and to the placement clause.

964    Moreover, it is settled that the exclusivity clause in GSAs concluded with all sites direct partners, the placement clause and the prior authorisation clause were able simultaneously to cover different parts of the market for online search advertising intermediation, it being specified that all GSAs containing the prior authorisation clause also contained the placement clause (see paragraphs 594 and 943 above).

965    Accordingly, as follows from paragraphs 603 and 857 above, the Commission cannot be considered to have committed an error of law  on the sole ground that it took into account the coverage of the exclusivity clause in GSAs concluded with all sites direct partners as well as that of the placement clause in order to determine whether the coverage rate of the prior authorisation clause had been sufficient to prevent Google’s competitors from accessing a significant part of the market at issue.

966    It follows that the Commission cannot be criticised for having taken into account GSAs concluded by direct partners that were never subject to the prior authorisation clause, or for having taken into account GSAs containing that clause, but in which direct partners had not typically included all of their websites.

967    Furthermore, it has been noted in paragraphs 631 and 859 above that the combined coverage rate of the exclusivity clause in GSAs in which direct partners had typically included all of their websites, on the one hand, and the placement clause, on the other hand, could be sufficient to be capable of producing a foreclosure effect between 2006 and 2015. It follows that the coverage rate of the prior authorisation clause could also be sufficient to be capable of producing such an effect.

968    In those conditions, it is appropriate to find that the circumstance, cited by Google, that a significant part of the online search advertising intermediation market was not covered by the prior authorisation clause is not such as to rule out that that clause was capable of producing a foreclosure effect.

969    However, it should be noted that the Commission did not determine the coverage rate of the prior authorisation clause in relation to 2016. It is limited to stating, in the contested decision, that Google had informed it, on 28 May 2016, that it was intending to remove that clause from all GSAs based on the March 2009 template GSA and that it had notified the last direct partner, on 6 September 2016, of its decision to waive the said clause. However, the Commission did not present, in that decision, evidence allowing for assessment of the extent of the coverage of GSAs containing the clause in question in relation to 2016. In that regard, it must be held that the Commission’s assertion formulated in the defence, according to which four direct partners were still subject to such GSAs until 3 June 2016 is not such as to call that finding into question.

970    Accordingly, it is appropriate to find that the Commission did not establish that the prior authorisation clause could have prevented Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA in 2016.

(2)    Website traffic and number of online queries of the websites covered by the prior authorisation clause

971    Google considers, in essence, that the Commission could not prove that the prior authorisation clause had prevented its competitors from accessing a significant part of the online search advertising intermediation market by relying on the Microsoft study and on the fact, alleged in recital 589 of the contested decision referring to Table 27 thereof, that direct partners’ websites constituted a large part of all online queries performed in the EEA.

972    The Commission disputes Google’s line of argument.

973    In that regard, as follows from paragraphs 872 and 875 above, the Commission could merely determine the coverage rate of the prior authorisation clause in order to assess the extent of that coverage. In those conditions, it is appropriate to find that the Commission could rely on the Microsoft study and on the data recorded in Table 27 of the contested decision as indications corroborating the assessment of the said rate.

(b)    As-efficient competitor test

974    In recital 611 of the contested decision, the Commission noted that the prior authorisation clause was capable of foreclosing a hypothetical competing intermediary as efficient as Google. First, the revenues generated by GSAs containing that clause, between 2011 and 2015, represented between [confidential] and [confidential]% of the market for online search advertising intermediation in the EEA. Second, the revenues generated by the placement clause and the exclusivity clause in GSAs concluded with all sites direct partners, between 2009 and 2015, represented [confidential] to [confidential]% of that market. Third, Google held a ‘very large’ share of the market between 2006 and 2016. Fourth, that market was prone to network effects.

975    The Commission moreover indicated, in recital 612 of the contested decision, that it was ‘doubtful’ whether a hypothetical intermediary as efficient as Google could have emerged during the period of application of the prior authorisation clause. It last found that the question of whether Google had pursued a strategy aimed at excluding competitors as efficient as it was irrelevant.

976    Google argues that the Commission failed to prove that a competitor as efficient as it could not have emerged on the market for online search advertising intermediation owing to the prior authorisation clause. It also criticises the Commission for having considered that the absence of a strategy aimed at competitors as efficient as it was irrelevant in this case.

977    The Commission disputes Google’s line of argument.

978    In that regard, it should be noted that, as has been mentioned in paragraphs 656, 657 and 880 above, the fact that the Commission indicated that it was ‘doubtful’ whether a hypothetical intermediary as efficient as Google could have emerged during the period of application of the prior authorisation clause is not such as to call into question the legality of the contested decision.

979    Moreover, as has been noted in paragraphs 665 and 881 above, the Commission could merely demonstrate the capability of the prior authorisation clause to produce a foreclosure effect by relying on several relevant elements, without necessarily relying, to that end, on the as-efficient competitor test. Furthermore, as has been noted in paragraph 671 above, it is settled that Google did not provide during the administrative procedure – or before the Court – any analysis based on that test.

980    In the case at hand, the very substance of the prior authorisation clause, which essentially provided that direct partners had to obtain Google’s written approval before changing the display of competing ads, and the factors mentioned in recital 611 of the contested decision, namely, on the one hand, the fact that the prior authorisation clause, together with the exclusivity clause in GSAs in which direct partners had typically included all of their websites and the placement clause, covered a significant part of the online search advertising intermediation market, as has been noted in paragraphs 650 and 859 above, and, on the other hand, the extent of Google’s dominant position resulting in particular from its very high market shares and from the barriers to entry and expansion in the form inter alia of network effects, could demonstrate that the prior authorisation clause could be capable of foreclosing a hypothetical competitor as efficient as Google. It follows that those factors could also demonstrate that it was ‘doubtful’ that such a competitor could have emerged during the period of application of the said clause.

981    Last, as follows from paragraphs 678 to 681 above, the Commission cannot be criticised, on the one hand, for not having established that Google had adopted a strategy aimed at excluding competitors at least as efficient as it and, on the other hand, for not having taken into account the fact that it did not intend to exclude such competitors.

(c)    Duration of GSAs and the unilateral termination right of direct partners

982    In the contested decision, the Commission considered that the average duration of GSAs containing the prior authorisation clause was long. It stated, in that regard, that Google and direct partners had extended certain GSAs, sometimes several times, without substantial modifications.

983    Google criticises the Commission for having confused the length of the commercial relationship with the direct partners subject to the prior authorisation clause with the duration of the GSAs containing that clause. It thus notes that the GSAs mentioned in footnotes 766 and 767 to the contested decision had a duration of two years or less between each renewal or each extension. In addition, it states that the Commission failed to take into account the fact that some direct partners held a unilateral termination right.

984    Vinden claims that the duration of the prior authorisation clause was short and that it had a unilateral termination right by virtue of its 2011 GSA.

985    The Commission refers to its line of argument relating to the merits of Google’s line of argument concerning the exclusivity and placement clauses. It disputes Google’s allegations that, first, the duration of each GSA between two renewals was two years or less and, second, some direct partners had unilateral termination rights.

986    In that regard, as has been indicated in paragraph 919 above, the Commission considered, in the contested decision, that the prior authorisation clause had, on the one hand, deterred direct partners from sourcing from Google’s competing intermediaries and had, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation.

987    Accordingly, it should be noted that the Commission considered – as it moreover indicated, in essence, in recitals 629 and 630 of the contested decision – that the prior authorisation clause produced a foreclosure effect similar to that of the exclusivity clause in GSAs concluded with all sites direct partners, on the one hand, and to that of the placement clause, on the other hand.

988    In those conditions, it must be held that, as follows from paragraphs 695, 696 and 891 above, the duration of the obligation by which direct partners had to request authorisation from Google before being able to change the display of competing ads counts among the circumstances relevant for assessing the foreclosure effect of that clause.

989    On the one hand, however, it should be noted that, in order to regard the average duration of GSAs containing the prior authorisation clause as long, the Commission relied only, in recital 590 of the contested decision, which refers to recital 519 thereof, mentioned in paragraph 892 above, and in recital 594 thereof, on the fact that certain GSAs had been extended, sometimes several times, without substantial modifications. More specifically, as follows from paragraph 893 above, it is apparent from Annex A.55 to the application and from footnotes 766 and 767 to the contested decision, the content of which restates that of footnote 713 thereto, that the Commission took into account the total duration of those GSAs, by including all of their extensions, where applicable. However, it took into account neither the initial duration of each of the said GSAs, taken individually, nor the duration of each of their extensions, where applicable.

990    On the other hand, it must be pointed out that, in its analysis of the effects of the prior authorisation clause, the Commission made no mention of the unilateral termination rights held by some of the direct partners subject to that clause. It is apparent from Annexes A.55 to the application and C.8 to the reply, however, that at least five direct partners mentioned in Table 28 of the contested decision, namely Vinden, [confidential], [confidential], [confidential] and [confidential], held a unilateral termination right, which the Commission does not dispute. Moreover, it is apparent from footnotes 766 and 767 to that decision that the Commission relied on the duration of GSAs concluded only with 11 direct partners. Accordingly, in line with what has been found in paragraph 897 above, the fact that at least five direct partners held a unilateral termination right cannot be regarded as a sufficiently rare exception to justify its not being taken into account for the purposes of the assessment of the foreclosure effect of the prior authorisation clause.

991    In those conditions, and as follows from paragraph 898 above, the Commission could not, solely on the basis of recitals 590 and 594 of the contested decision and without having examined the actual conditions and the terms under which GSA extensions had been agreed, as well as the substance of the clauses providing for the unilateral termination rights held by some of the direct partners and the conditions in which those rights could be exercised, exclude that those direct partners had the option of sourcing from Google’s competing intermediaries without having to request prior authorisation from it to change the display of competing ads, including before any extension of their GSAs, or before a unilateral termination right had been exercised. It follows that, in those conditions, the Commission also could not find that those intermediaries had not had the possibility of disputing the part of the market for online search advertising intermediation covered by GSAs containing the prior authorisation clause for the total duration of those GSAs, let alone for their cumulative duration.

(d)    Conclusion on the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

992    As has been found in paragraph 967 above, the Commission was right to consider that, having regard to the coverage of the exclusivity clause in GSAs concluded by all sites direct partners and of that of the placement clause, in the light of the circumstances recalled in paragraph 602 above, the coverage of the prior authorisation clause could be sufficient to enable that clause to be capable of producing a foreclosure effect between 31 March 2009 and 31 December 2015. In addition, it has been noted in paragraph 980 above that Google was not justified in arguing that the Commission had failed to prove that a competitor as efficient as it could not emerge on the market for online search advertising intermediation or that such a competitor was capable of being excluded from that market.

993    However, on the one hand, as follows from paragraph 969 above, the Commission did not establish that the prior authorisation clause could have produced a foreclosure effect, owing to its coverage, between 1 January and 6 September 2016. On the other hand, as follows from paragraph 991 above, it failed to take into consideration all the relevant circumstances of the case in the context of the assessment of the duration for which direct partners were obliged, by virtue of that clause, to request prior authorisation from Google before changing the display of competing ads on their results pages.

994    It follows that, contrary to what is required by the case-law recalled in paragraph 107 above, the Commission has not demonstrated to the requisite legal standard, in the light of all the circumstances of the case, that the prior authorisation clause was capable of preventing Google’s competing intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA, for the duration for which that clause applied.

3.      Conclusion on the first part of the fourth plea

995    As has been recalled in paragraph 919 above, the Commission considered, in recital 573 of the contested decision, that the prior authorisation clause was capable of restricting competition, in the light of all the circumstances relevant to the case. It noted in that regard that that clause had (i) deterred direct partners from sourcing from Google’s competing intermediaries, (ii) prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation, (iii) possibly deterred innovation, (iv) helped Google to maintain and strengthen its dominant position on the national markets for online search advertising in the EEA, with the exception of Portugal, and (v) possibly harmed consumers.

996    As has been recalled in paragraphs 920 and 921 above, the Commission essentially considered that the prior authorisation clause was capable of producing a foreclosure effect, finding that it had, on the one hand, deterred direct partners from sourcing from Google’s competing intermediaries and had, on the other hand, prevented those intermediaries from accessing a significant part of the market for online search advertising intermediation. In addition, it inferred from that foreclosure effect that that clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

997    As has been recalled in paragraph 926 above, in recital 574 of the contested decision, before examining the effects of each of the five restrictions of competition that it identified (see paragraph 995 above), the Commission specified that, for the purposes of its analysis seeking to demonstrate that the prior authorisation clause was capable of restricting competition, it had taken into account the ‘duration of [that c]lause’, referring, in that regard, to Section 8.5.4.2 thereof, relating to the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation, and thus highlighting, rightly, the importance of the developments set out in recitals 589 and 594 included in that section. It also indicated that it had taken into consideration the coverage rate of the said clause, which it examined in the same section of the said decision. It is also apparent from the systemic place of the said recital 574 in the structure of the contested decision that the Commission took into account that duration and that coverage rate when it examined the effects of the prior authorisation clause in the context of each of the five restrictions identified in that decision.

998    As has been found in paragraph 993 above, however, the Commission failed to take into consideration all the relevant circumstances of the case in the context of the assessment of the duration for which direct partners had been obliged, by virtue of the prior authorisation clause, to request prior authorisation from Google before changing the display of competing ads on their results pages.

999    Moreover, as has also been found in paragraph 993 above, the Commission did not establish that the prior authorisation clause could have produced a foreclosure effect, owing to its coverage, between 1 January and 6 September 2016.

1000 It follows that the errors committed by the Commission, recalled in paragraphs 998 and 999 above, vitiate all of the restrictions identified by it in the contested decision, such that it is appropriate to conclude that the Commission has not demonstrated, to the requisite legal standard, that the prior authorisation clause had been capable of deterring direct partners from sourcing from Google’s competing intermediaries or that it had been capable of preventing those intermediaries from accessing a significant part of the market for online search advertising intermediation in the EEA and, consequently, that that clause had been capable of having the foreclosure effect found in the said decision.

1001 In those conditions, it is appropriate to find that the Commission has also not demonstrated, to the requisite legal standard, that the prior authorisation clause had, first, possibly deterred innovation, next, helped Google to maintain and strengthen its dominant position on the national markets for online search advertising at issue and, last, possibly harmed consumers.

1002 It follows that the first part of the fourth plea must be upheld, without it being necessary to rule on the other arguments of Google under that plea, and, consequently, that the contested decision must be annulled to the extent that it found that the prior authorisation clause constituted an infringement of Article 102 TFEU.

F.      Conclusion on the action

1003 It follows from paragraphs 727, 913 and 1002 above that the Commission has not established any of the three infringements of Article 102 TFEU constituting the single and continuous infringement of that same provision, mentioned in Articles 1 to 3 of the contested decision. Moreover, as Google essentially argues, it is apparent from the scheme and operative part of that decision that the Commission considered that that single and continuous infringement was characterised only in so far as it consisted of separate infringements.

1004 Accordingly, the contested decision must be annulled to the extent that it found that the exclusivity clause in GSAs concluded with all sites direct partners, the placement clause and the prior authorisation clause constituted, together, a single and continuous infringement of Article 102 TFEU, without it being necessary to rule on the merits of Google’s line of argument aimed at challenging specifically the classification of such a single and continuous infringement and, consequently, on the plea of inadmissibility, raised by the Commission, alleging that that line of argument is inadmissible.

1005 It follows from all of the foregoing that the contested decision must be annulled in its entirety, without there being any need to rule on Google’s fifth plea in law.

IV.    Costs

1006 First, under Article 134(1) of the Rules of Procedure, the unsuccessful party is to be ordered to pay the costs if they have been applied for in the successful party’s pleadings. Since the Commission has been unsuccessful, it must be ordered to pay the costs incurred by Google, in accordance with the form of order sought by the latter.

1007 Second, in accordance with Article 138(3) of the Rules of Procedure, the Court may order an intervener other than those referred to in paragraphs 1 and 2 of that article to bear its own costs. In this case, Surfboard and Vinden must be ordered to bear their own costs.

On those grounds,

THE GENERAL COURT (Tenth Chamber, Extended Composition)

hereby:

1.      Annuls Commission Decision C(2019) 2173 final of 20 March 2019 relating to a proceeding under Article 102 TFEU and Article 54 of the EEA Agreement (Case AT.40411 – Google Search (AdSense));

2.      Orders the European Commission to pay the costs of Google LLC and of Alphabet Inc.;

3.      Orders Surfboard Holding BV and Vinden.NL BV to bear their own costs.

Delivered in open court in Luxembourg on 18 September 2024.

Table of contents

I. Background to the dispute

A. Google’s online search advertising intermediation services and contracts

B. Administrative procedure

C. The contested decision

1. Market definition

(a) National markets for online search advertising

(1) Product market

(2) Geographic market

(b) Market for online search advertising intermediation in the EEA

(1) Product market

(2) Geographic market

2. Dominant position

(a) National markets for online search advertising

(1) Market shares

(2) Barriers to entry and expansion

(3) Lack of countervailing buyer power

(b) Market for online search advertising intermediation in the EEA

(1) Market shares

(2) Barriers to entry and expansion

(3) Lack of countervailing buyer power

3. Exclusivity clause in GSAs in which direct partners had ‘typically’ included all of their websites

(a) Exclusivity clause in GSAs concluded with all sites direct partners to the extent that it constituted an exclusive supply obligation contrary, as such, to Article 102 TFEU

(b) Exclusivity clause in GSAs concluded with all sites direct partners to the extent that it constituted an exclusive supply obligation capable of restricting competition within the meaning of Article 102 TFEU

(c) Absence of objective justification

4. Placement clause

(a) Scope of the placement clause

(b) Restriction of competition as a result of the placement clause

(c) Absence of objective justification

5. Prior authorisation clause

(a) Restriction of competition as a result of the prior authorisation clause

(b) Absence of objective justification

6. Single and continuous infringement

7. Effect on trade between Member States

8. The fine

II. Forms of order sought

III. Law

A. Preliminary observations

B. First plea: the Commission erroneously defined the relevant markets at issue and Google’s dominant position

1. First part of the first plea: substitutability between online search ads and online non-search ads

(a) Preliminary observations

(b) Taking into account of the perspective of publishers

(c) Taking into account of all relevant factors

(d) Google’s arguments relating to the SSNIP test

(1) Adequacy of the price analysis carried out by the Commission

(2) Merits of the conclusions drawn by the Commission from the price analysis that it carried out

(i) Interpretation of the replies to the question on prices

– Interpretation of the replies of publishers

– Interpretation of the replies of advertisers

– Interpretation of the replies of media agencies

– Conclusion on the Commission’s interpretation of the replies to the question on prices

(ii) Alleged misrepresentation of the replies to the question on prices

(e) Merits of the Commission’s analysis concerning the differences in characteristics between the two types of ad at issue

(1) Triggering and positioning of the two types of ad at issue

(2) Formats of the two types of ad at issue

(3) Design costs of the two types of ad at issue

(4) Targeting abilities of the two types of ad at issue

(5) Purposes of the two types of ad at issue

(6) CTRs and conversion rates of the two types of ad at issue

(7) Possibilities of measuring the performance of the two types of ad at issue

(8) Relevance of the differences in characteristics and uses to the definition of the market

(f) Taking into account of examples of the actual conduct of publishers which have allegedly replaced or would replace online search ads with online non-search ads

(g) Interpretation of the statements of certain Google representatives

(h) Past Commission decisions

(i) Conclusion on the first part of the first plea

2. Second part of the first plea: substitutability of the sale of online ads via an intermediary and the sale of such ads directly by publishers

(a) Substitutability of the two sales channels at issue from the perspective of publishers

(1) Transaction costs for publishers

(2) Indicator relating to access to a sufficient advertiser base

(3) Lack of ‘proper’ analysis of the substitutability of the two sales channels at issue

(4) Publishers using the two sales channels at issue

(b) Substitutability of the two sales channels at issue from the perspective of advertisers

(c) Taking into account of the Telefónica UK decision

(d) Conclusion on the second part of the first plea

3. Conclusion on the first plea

C. Second plea: the exclusivity clause in GSAs concluded with all sites direct partners did not constitute an abuse of a dominant position

1. First and second parts of the second plea: the exclusivity clause in GSAs concluded with all sites direct partners did not constitute an exclusive supply obligation within the meaning of the case-law resulting from the judgment of 13 February 1979, Hoffmann-La Roche v Commission (85/76, EU:C:1979:36) and the Commission was required to analyse the effects of that clause

2. Third part of the second plea: the contested decision does not establish that the exclusivity clause in GSAs concluded with all sites direct partners was capable of restricting competition

(a) Deterrent effect of the exclusivity clause vis-à-vis all sites direct partners

(1) Choice of direct partners to include a website in a GSA

(2) Replies of direct partners to the Commission’s various requests for information and Surfboard’s letter

(i) Relevance of the replies to Question 5.2.d of the request for information of 22 December 2010

(ii) Relevance of the replies of direct partners that are not all sites direct partners

(iii) All sites direct partners identified in the contested decision

– Relevance of Google’s line of argument

– Reliability of the replies to the request for information of 24 February 2017

(iv) Replies of the all sites direct partners mentioned in recitals 367 and 368 of the contested decision

– Replies of [confidential]

– Replies of the [confidential] group

– Replies of [confidential]

– Replies of the [confidential] group

– Replies of the [confidential] group

– Replies of the [confidential] group, [confidential] and the [confidential] group

(v) Other replies of the all sites direct partners cited by Google

– Replies of [confidential] and [confidential], [confidential], [confidential], the [confidential] group and the [confidential] group

– Replies of [confidential] and the [confidential] group

(vi) Surfboard’s letter

(vii) Conclusion on the replies of direct partners to the Commission’s various requests for information and Surfboard’s letter

(3) Investments made by Yahoo!

(4) Preference of publishers for procuring online search ads from one intermediary at a time

(i) Study produced by Google during the administrative procedure

(ii) Replies of direct partners cited by Google

(5) Conclusion on the deterrent effect of the exclusivity clause vis-à-vis all sites direct partners

(b) Impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

(1) Application of the exclusivity clause to certain online search ad formats

(2) Coverage of the market by the exclusivity clause in GSAs concluded with all sites direct partners

(i) Taking into account of data subsequent to the conclusion of GSAs for the purpose of calculating the coverage rate of the exclusivity clause

(ii) Taking into account of the revenues generated by GSAs concluded with direct partners belonging to the same group as certain all sites direct partners

(iii) Taking into account of GSAs containing placement and prior authorisation clauses

– Taking into account of the revenues generated by GSAs containing the placement and prior authorisation clauses for the purposes of assessing the foreclosure effects of the exclusivity clause in GSAs concluded with all sites direct partners

– Taking into account of the revenues generated by GSAs containing the placement and prior authorisation clauses concluded with direct partners that typically did not include all of their websites in those GSAs for the purposes of assessing the foreclosure effects of the exclusivity clause in GSAs concluded with all sites direct partners

(iv) Taking into account of direct partners that had not included all of their websites in their GSAs containing the exclusivity clause for the purpose of calculating the coverage rate of that clause

(v) Revenues generated by GSAs concluded with all sites direct partners in 2016

– Exclusivity clause in GSAs concluded with all sites direct partners

– Placement clause

(vi) Traffic of the websites covered by the exclusivity clause in GSAs concluded with all sites direct partners

(vii) Conclusion on the coverage of the market by the exclusivity clause in GSAs concluded by all sites direct partners

(3) As-efficient competitor test

(i) Preliminary observations

(ii) Factors relevant to the application of the as-effective competitor test

(iii) Evidence submitted by Google during the administrative procedure

(iv) Existence of a strategy aimed at excluding as-efficient competitors

(v) Conclusion on the as-efficient competitor test

(4) Duration of GSAs and the unilateral termination right of certain direct partners

(i) Admissibility of Google’s line of argument

(ii) Merits of Google’s line of argument

(5) Conclusion on the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

(c) Conclusion on the third part of the second plea

D. Third plea: the placement clause did not constitute an abuse of a dominant position

1. First part of the third plea: misinterpretation of the scope of the placement clause

(a) Possibility of showing competing ads below Google ads

(b) Spaces generating the highest CTR

(1) Illustrations in Annex A.52 to the application

(2) Figures 5 and 6 of Annex C.11 to the reply

(c) Conclusion on the first part of the third plea

2. Second part of the third plea: no restriction of competition as a result of the placement clause

(a) Deterrent effect of the placement clause vis-à-vis direct partners

(1) Capability of the placement clause to prevent direct partners from evaluating the commercial interest in sourcing from Google’s competing intermediaries

(i) Scope of the placement clause

(ii) Statements of direct partners

– Admissibility of Google’s line of argument

– Merits of Google’s line of argument

(2) Capability of the placement clause to prevent direct partners from being able to adopt certain configurations for their results pages, when the user visited those pages from a desktop device

(3) Commercial interest in direct partners in sourcing from Google’s competing intermediaries in the absence of the placement clause

(4) Conclusion on the deterrent effect of the placement clause vis-à-vis direct partners

(b) Impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

(1) Application of the placement clause to certain online search ad formats

(2) Market coverage by the placement clause

(i) Market share covered by the placement clause

(ii) Traffic and number of online queries of the websites covered by the placement clause

(3) As-efficient competitor test

(4) Duration of GSAs and the unilateral termination right of some direct partners

(5) Conclusion on the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

(c) Conclusion on the second part of the third plea

E. Fourth plea: the prior authorisation clause did not constitute an abuse of a dominant position

1. Deterrent effect of the prior authorisation clause vis-à-vis direct partners

2. Impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

(a) Market coverage by the prior authorisation clause

(1) Market share covered by the prior authorisation clause

(2) Website traffic and number of online queries of the websites covered by the prior authorisation clause

(b) As-efficient competitor test

(c) Duration of GSAs and the unilateral termination right of direct partners

(d) Conclusion on the impossibility for Google’s competing intermediaries to access a significant part of the market for online search advertising intermediation

3. Conclusion on the first part of the fourth plea

F. Conclusion on the action

IV. Costs

*      Language of the case: English.

1 Confidential information redacted.

© European Union
The source of this judgment is the Europa web site. The information on this site is subject to a information found here: Important legal notice. This electronic version is not authentic and is subject to amendment.