Introduction
The decision of the ECJ of 10th of September 2024 (C-48/22 P) brings the Google Shopping saga to an end. The decision by the highest court confirms the conviction of Alphabet by the General Court (T-612/17) (see the blogpost on General Court’s decision here). The European Commission had previously imposed a record fine of 2.42 billion euros on Alphabet (AT.39740) for favouring its own price comparison service and the General Court essentially confirmed the decision in an equally eagerly awaited ruling of 10th of November 2021.
The so-called self-preferencing by large digital platforms is currently the focus of antitrust authorities, antitrust law scholars and legislation and will continue to be one of the defining topics in the field of competition and antitrust law in the coming years. It involves a company exploiting its prominent position in a market (usually a platform market) to gain advantages over his competitors in a related downstream market. Many competition authorities around the world are taking action against this behaviour and are increasingly opening proceedings against the “internet giants”. First and foremost, the European Commission, which, in addition to the Google Shopping case, is also bringing or have recently brought further proceedings against Amazon (Marketplace seller data, Buy box and Prime, under DMA) Apple (App Store Rules here, here and here, under DMA), Meta (under DMA), Alphabet (Google Android, under DMA) and Microsoft (Microsoft Cloud), etc. regarding self-preferencing practices.
In the Google Shopping case, the European Commission accused Alphabet (Google’s parent company) of artificially favouring its own price comparison service “Shopping” over other price comparison services and therefore found an abuse of a dominant market position under Art. 102 TFEU and imposed the record fine of 2.42 billion euros upon Alphabet. Google’s price comparison service had been favoured by being placed prominently at the top of the search results. At the same time, price comparison services from other providers such as Yelp, Idealo.de, etc. were downgraded in the generic search results and only displayed on page four on average. We know from statistics, that users very often click on the first search results and only a very small number of clicks are made on search results on the second page and thereafter. However, it has not yet been scientifically clarified whether this is the use of a non-performance-related means or merely the exploitation of an advantageous position gained through competition on the merits. At the same time, developments at regulatory level have come thick and fast. The German legislator reacted with the introduction of the new Section 19a (2) sentence 1 no. 1 ARC and the European legislator joined in by the implementation of the self-preferencing prohibition pursuant to Art. 6 (5) DMA. In the meantime, similar legislative measures can also be found in the US ( Sec. 3 lit. a (1) American Innovation and Choice Act here and Sec. 3 lit. e Open App Markets Act here).
With this decision, the ECJ addresses the intriguing question of whether or not self-preferencing in this or similar cases constitutes behaviour by a dominant company that is based on competition on the merits and is therefore compliant under antitrust law. The decisive factor is the effects of the conduct, i.e. whether the practice is likely to have negative effects on competition. In this context, it is open to question whether such cases do involve a refusal to grant access to an essential facility, for which the criteria set out in the Bronner case must be taken into account when examining the compliance with antitrust law. In view of the new provisions in Section 19a ARC and Art. 6 (5) DMA, it is noteworthy that the first case concerning self-preferencing by digital platforms is assessed on the basis of Art. 102 TFEU and thus also sheds light on the new legislative measures
Bronner criteria
The question of the application of the Bronner criteria in this case was particularly very controversial in literature and practice.
This is because the preferred presentation of the company’s own price comparison service consisted of so-called One Boxes, which were placed above the generic search results in a graphically more appealing way with images and prices. At the same time, the boxes has not been subject to the algorithm for the generic search, which is why certain negative criteria for the ranking of price comparison tools did not apply to Google’s own price comparison service.
Google therefore argues that these boxes, which can be distinguished from the generic search results in terms of graphics and content, are considered as a separate infrastructure to which the price comparison services of other providers have not had access yet, but which were essential for their business model. These were clearly more desirable for the competing price comparison services, as the Court also found (para. 71).
Facts of the case
This case is somewhat similar to the situation in the Bronner case (ECJ, judgement of 26 November 1998). As a reminder: Oscar Bronner GmbH & Co. KG, which publishes the daily newspaper “Der Standard” in Austria with a market share of 3.6% and was distributed in Austria mainly via newsagents and postal delivery, demands access to Mediaprint’s home delivery distribution system (for a fee). There are several home delivery networks for daily newspapers in Austria, but only at regional or local level. The only nationwide home delivery system is operated by the Austrian Mediaprint newspaper and magazine publisher, which publishes the “Neue Kronen Zeitung” among others, via its wholly-owned subsidiary Mediaprint Zeitungsvertriebsgesellschaft. The ECJ denied an abuse of market power pursuant to Art. 102 TFEU by Mediaprint refusing access to its own in-house delivery system, as the criteria of the essential facilities doctrine were not met. Accordingly, it must be an essential facility for entering on another market access to which is not granted and the refusal cannot be objectively justified by weighing the dominant company’s proprietary interest with the public interest. Finally, the refusal of access must be capable of excluding all competition on the derived market. The ECJ rejected Oscar Bronner‘s claim for access, because a home delivery system already lacks an essential facility for operating on a derived market. This is because there are other suitable distribution channels in the form of postal delivery and shop or kiosk sales, whose disadvantages for the distribution of daily newspapers do not exclude the activity as such. Furthermore, there are no technical, legal or economic obstacles for a publisher to setting up a nationwide home delivery system on one’s own or in co-operation with other publishers. It is not sufficient that the creation of such a system is unprofitable due to the low circulation of the newspaper or newspapers to be distributed. Instead, the creation of the home delivery system would have to be unprofitable precisely for the distribution of daily newspapers with a circulation comparable to that of the existing system. The ECJ therefore denied an abuse of market power by Mediaprint refusing access of Bronner to its home delivery system.
Alphabet argued in its defence in the Google Shopping case in a similar way. The Google search engine did not grant competing price comparison services equal access to the more attractive display option, the boxes of the Shopping Units, which is better than the presentation within the generic search results in terms of positioning, features and lack of risk of downgrading. These boxes were therefore to be qualified as an essential facility for competitors to take up their activities on the market for price comparison services, to which they had not been granted access (para. 71 ECJ decision). This would already follow from the General Court’s decision, which itself states that the boxes provide their own technical infrastructure and it is therefore irrelevant that access was granted to the other display option, the generic search results.
Application of the Bronner criteria
Essential facility: Shopping unit boxes?
However, the ECJ, like the General Court and the Advocate General, did not pursue this distinction. This is to be welcomed, as a distinction between boxes and generic search results as different infrastructures to which separate access could be granted would be artificial. Although the boxes are prominently displayed on the general results page, they do not represent a separate infrastructure in the sense of a technically completely independent results page. In contrast to its formerly independent predecessor Froogle product search service, Shopping has been objectively and subjectively integrated into the general search in order to utilise the network effects generated there to its own advantage, i.e. it is part of it. The shopping units are therefore a means of presentation within the search results, serve the same purpose, namely to facilitate the exchange of information between providers and searchers, and are ultimately not to be regarded as a separate infrastructure. From the end user’s perspective, the Google search results pages in their entirety also appear to be uniform in terms of usability and overall presentation, (including the Shopping Units) are initiated by a search request in the general search and a difference only becomes recognisable at second glance. However, this is precisely intentional, as network effects have a significantly lower impact when own services are recognised as being preferred, because users then no longer automatically attribute higher quality to the ranking position. As Advocate General Kokott pointed out in point 115 of her opinion, the discrimination against the search results of competing price comparison services therefore concerns the form and manner of access to Google’s general results page, but is not a question of access to an allegedly separate infrastructure in the form of the boxes of the Shopping Units. The competitors have credibly argued that they never requested access to these boxes, but rather demanded their abolition.
Google, on the other hand, already granted competing price comparison services access to the generic search results – in line with the logic of its business model – because it offers a fundamentally open infrastructure designed to attract a maximum number of internet users and generate traffic in order to achieve the positive network effects necessary for its commercial success. (paras. 177, 178 GC decision).
The GC’s statement that the general results page has characteristics “akin to those of an essential facility” is particularly unfortunate in this context, as the court thereby provoked Google’s objection to taking the Bronner criteria into account, or at least served it up on a silver platter. However, this was only to be seen as an independent assessment of the Commission’s essential justification for the abuse of market power, as the Commission had found that there was no economically viable alternative for the competing price comparison services to the data traffic originating from Google’s general results page (paras. 224 and 225-227 GC decision). The “indispensability” mentioned by the court in para. 227 is not to be confused with the criterion of indispensability in the sense of the Bronner criteria – these are explicitly not applicable. Rather, according to the case law, this criterion is also part of the examination of abusive conduct by setting inappropriate delivery conditions, as AG Kokott explains in para. 89: Such practices may therefore be independent forms of abuse provided that they give rise to at least potentially anticompetitive effects, or even exclusionary effects, on the relevant market. That is all the more likely to be the case where access to such infrastructure, to a particular service or to a particular input is indispensable to allowing competitors of the dominant undertaking to operate profitably on a downstream market.“
The self-preferencing of which Google is accused is an independent form of abuse by applying inappropriate access conditions for competing price comparison services, provided that it has at least potential anti-competitive effects. The Bronner criteria are not applicable to such a form of abuse
Analogue application of the Bronner criteria
The Bronner criteria take into account the need to protect the fundamental right of the dominant undertaking to choose its trading partners and the freedom to dispose of its property, and any obligation of access would have to be regarded as an interference with those rights (para. 86 AG Kokott’s Opinion). A strict standard also serves to protect the incentives to innovate, because only if a company is allowed to use the facilities or (intellectual) property rights it has developed exclusively, the incentive to invest in them will be maintained. This incentive is a key driver of investment and innovation, which is precisely what the competitive process is supposed to promote. Less stringent criteria could, for their part, impair competition, also to the detriment of consumers. However, an imposed access or supply obligation could reduce or even eliminate these incentives for both the dominant company and its competitors. This is because it would allow competitors to participate in the fruits of such investments or innovations by the dominant company without having to invest in the development of a competing infrastructure themselves. This behaviour, also known as free riding, can therefore harm competition in the long term, including consumer welfare (para. 87 AG Kokott’s Opinion). In view of this ratio of the Bronner criteria, discrimination by self-preferencing should not be subject to such strict criteria. This would also unduly restrict the practical effectiveness of Art. 102 TFEU.
The ECJ confirmed the General Court’s view that not every case constellation that relates in part or solely to questions of access must necessarily be resolved by applying the criteria from the Bronner decision (para. 80-82 ECJ decision). Although the behaviour has similarities with the access issue, it differs from the refusal to supply in fundamental respects. According to the General Court, a refusal of access within the meaning of Bronner would require (1) an express refusal of access following an equally express request (see below) and (2) that the behaviour at issue consists mainly of the refusal as such and not of other conduct outside this framework, such as another form of leveraging abuse. This is because all or at least most foreclosure practices involve at least an implicit refusal to deal. Nevertheless, the Bronner criteria could of course not be applied to all such practices, as this would be contrary to “the spirit and the letter of Article 102 TFEU, the scope of which is not limited to abusive practices relating to ‘indispensable’ goods and services within the meaning of that judgment”. These are, for example, practices which, although they may ultimately amount to an implicit refusal to deal, constitute an independent infringement of Art. 102 TFEU because of their essential characteristics deviating from competition on the merits. In a number of cases in which problems of access to a service were at issue, such as margin squeeze cases, proof of indispensability was not required. This is because such practices can constitute a separate form of abuse. In Google Shopping the conduct does not merely consist of a unilateral refusal to supply, but rather of an active positive unequal treatment (para. 237 – 241 GC decision).
Art. 102 TFEU, as the ECJ points out, aims to prevent restrictions of competition to the detriment of the public interest, individual companies or individual consumers. The concept of competition on the merits must be used to draw a distinction, i.e. abusive behaviour occurs when means other than those of competition on the merits are used.
The Bronner criteria stipulate that the denial of access to an essential infrastructure facility only constitutes an abuse of market power if it would be capable of eliminating all competition on the downstream market, cannot be objectively justified and the infrastructure facility itself is indispensable for the performance of the competitor’s activity in the sense that there is no actual or potential substitute for it. Only in such cases is the serious interference with the freedom of contract and the right of property of the company, which solely developed the property itself, justified.
The boxes of the shopping units are, as the ECJ confirms, not a separate infrastructure facility (para. 105 et seq. ECJ decision). The relevant infrastructure is the results pages of the generic search results, to which the competing price comparison services have already been granted access. Google’s conduct therefore constitutes a form of discrimination, which is an independent form of abuse of market power through leveraging, which can be distinguished from a mere refusal of access (para. 109 et seq. ECJ decision). This is why there is also no space for indirect application of the bronner criteria.
Explicit request for access and refusal; active and passive behaviour as additional requirements of the bronner criteria
The General Court also rejected the application of the Bronner criteria because it assumes that a case of access only exists if there is an explicit request for access that has also been explicitly rejected. Google considers this very formalistic test to be contradictory in the context of established European case law and the legal and economic objectives of the essential facilities doctrine (para. 76 ECJ decision). However, the ECJ has not rejected the substance of this argument, but did not confirm it either. Since the General Court’s argument was not essential for the legally sound result – although the court could have found guilty providing additional justification in this respect compared to the Commission’s decision – Google’s complaint was rejected in the end.
The ECJ also left open the further reasoning of the court for a distinction criterion concerning the refusal of access in the sense of Bronner according to the activity or passivity of the conduct. The General Court argued that Google’s behaviour was more of an active, positive discrimination, whereas the refusal of access would be rather passive in nature. As the non-application of the Bronner criteria had already been clarified, this was no longer relevant for the decission and the ECJ also rejected the complaint against this reasoning of the General Court without taking position on the content (para. 117 ECJ decision).
The remedies as a classification criterion?
Google also argued that the remedies must be considered as the exact opposite of the alleged abuse of market power and thus indirectly characterised the type of abuse and therefore the test to be applied. However, the GC consistently rejected such a (mandatory) connection between the available remedies and the applicable legal test for the type of abusive conduct, as the ECJ confirmed.
“It explained that the obligation for an undertaking which is abusively exploiting a dominant position to transfer assets, enter into agreements or give access to its service under non-discriminatory conditions does not necessarily involve the application of the conditions set out in the judgment of 26 November 1998, Bronner (C‑7/97, EU:C:1998:569). There could, after all, be no automatic link between the criteria for the legal classification of the abuse and the corrective measures enabling it to be remedied. The General Court added that the fact that one of the ways of ending the abusive conduct was to allow competitors to appear in the boxes displayed at the top of the Google results page did not mean that the abusive practices should be limited to the display of those boxes and the conditions for identifying the abuse should be defined having regard to that aspect alone.”
(para. 86 ECJ decision)
Competition of the Merits
The second question in the Google Shopping case, which was hotly debated in the run-up, was whether taking advantage of an organically grown position of market power to favour own products and services is inherently consistent with competition on the merits or not. With the second ground of appeal, Alphabet therefore challenged the General Court’s assessment that the preferential treatment of Google’s own price comparison service was anti-competitive and deviated from competition on the merits.
Specific circumstances
In its grounds of appeal, Google referred to the three specific circumstances identified by the General Court on which the assessment as “falls outside of the scope of competition on the merits” was based. In doing so, the appeal accused the GC of incorrectly categorising the conduct in legal terms on the basis of the effects and their factual circumstances and of mixing them up. The three circumstances at issue are (1) the importance of the traffic from Google’s general search engine to the price comparison services, (2) the user behaviour when searching online and (3) the fact that the redirected traffic from Google’s general results pages accounted for a large proportion of the traffic to the competing price comparison services and could not be effectively replaced by other sources.
The ECJ dealt with this question in more detail than the Advocate General in her opinion and referred to the origin of the concept of competition on the merits:
In order to assess the ground for appeal, it should be recalled that Art. 102 TFEU, even if it imposes a special responsibility on dominant undertakings for their competitive behaviour in order to protect undistorted competition, does not prohibit the existence of a dominant position itself, but only its abusive exploitation (para. 163 ECJ decision). Art. 102 TFEU is neither intended to prevent companies from achieving a dominant position on one or more markets through their own performance (so-called competition on the merits), nor to ensure that competitors that are less efficient than the dominant company remain on the market (so-called as-efficient competitor). On the contrary, competition on the merits can, by definition, lead to less efficient competitors being forced out of the market (para. 164 ECJ decision). According to case law, the leverage effects applied by a dominant company as such are not prohibited per se under Art. 102 TFEU too, but it can only be declared after a case-by-case examination of all relevant specific circumstances whether or not this behaviour is in line with competition on the merits.
Since Google had accused the General Court and the Commission of confusing the legal classification of the self-preferencing strategy with its effects, it should be recalled that evidence of actual or potential anti-competitive effects is also required in the context of abuse of market power under antitrust law and the labelling of the conduct in question (para. 165 ECJ decision). These may be demonstrated not only by the crowding out of actual competitors, but also by hindering of potential competitors from entering the market. All relevant factors of the individual case, in addition to the practice in question, the relevant market (or markets) as well as the functioning of competition on this market(s), must be taken into account in this assessment. The conduct in question must at least be capable of giving rise to a crowding-out effect, which can be demonstrated by various methods of analysis.
Against this background, the Commission’s approach cannot be criticised. The antitrust authority determined the antitrust infringement pursuant to Art. 102 TFEU on the basis of the specific circumstances of the individual case and was not content with merely identifying a leverage effect or inferring this merely from the existence of foreclosure effects, but rather legally classified Google’s (self-favoring) practices. The Commission based its assessment legally on the relevant criteria and thus found without error of law that the self-preferencing of Google’s price comparison service was not in line with means associated with competition on the merits. The ECJ confirmed that the factors identified by the General Court constitute relevant elements of the context which are capable of characterising conduct which is not in conformity with competition on the merits (para. 168 ECJ decision). The court was therefore allowed to use these factors as relevant for the classification of the behaviour. Only concerning these factors it was possible to place the two-stage behaviour in the context of the two markets concerned and the functioning of competition on these markets. The circumstances were such as to demonstrate that the success of Google’s price comparison service on the downstream market and the potential exclusionary effects were not based on conduct consistent with competition on the merits, but were based on the practices at issue in conjunction with the specific circumstances established (para. 171 ECJ judgment).
And, as AG Kokott indicated in point 143 of the opinion, the General Court did not in any way confuse the analysis of that conduct with the analysis of its effects when examining the question of whether the conduct at issue deviated from competition on the merits.
The “superdominant” company as a new target group and the assessment of discrimination
In the second ground of appeal, Google further challenged the court’s decision, alleging that it had illegally applied additional considerations, i.e. introduced a stricter legal test for so-called “superdominant” companies, that the self-preferencing was abnormal in terms of competition and that the conduct was discriminatory. Google argued that the court had introduced a stricter legal test for “super-dominant” companies, although the law previously only recognised dominant and non-dominant companies. However, this concern was misplaced. Both the Advocate General and the ECJ found that, although the court had actually introduced this concept for the first time in this proceeding, it had only done so for the sake of completeness and the considerations were therefore not necessary for the result.
In addition, Google accused the General Court of having erred in law by introducing new reasoning in the context of the assessment of discrimination and for avoiding to examine the existence of arbitrariness. In fact, the General Court had only summarised the Commission’s statements and confirmed that the practices at issue, which manifested themselves in the form of positive discrimination in the treatment, were an independent form of leveraging abuse arising from a dominated market characterised by high barriers to entry, namely the market for general search services.
It was interesting that the ECJ in this context referred to a principle, namely “that it cannot be considered that, as a general rule, a dominant undertaking which treats its own products or services more favourably than it treats those of its competitors is engaging in conduct which departs from competition on the merits irrespective of the circumstances of the case” (para. 186 ECJ decision). In the present case, the General Court confirmed the Commission’s analysis that the conduct in question, with its two elements, i.e. the highlighting of its own results and the downgrading of the results of competing services, was discriminatory in view of the characteristics of the upstream market and the specific circumstances found and did not correspond to competition on the merits. The discrimination was therefore based on the origin of the results of the price comparison services and had an impact on presentation and positioning (para. 183 ECJ decision). The practice was realised through the exploitation of a leverage effect and was therefore capable of weakening competition on the market for price comparison services and was consequently classified as conduct falling outside the scope of competition on the merits. Google was thus able to exploit its dominant position on the upstream market for general online search services, in order to gain competitive advantages on the downstream, not yet dominated market for specialised search services, namely by favouring its own price comparison service.
Google’s objection that it was not even able to display the specialised search results of competing price comparison services with the same reliability, presentation and positioning as its own price comparison service had already been rejected as unfounded by the lower court.
Fun fact: Even highly qualified judges like those of the General Court happen to make a typing mistake in an official document (para. 575 GC decision). Fortunately, it did not affect the outcome of the decision 😉
Counterfactual analysis
Briefly, as regards the rejection of the third ground of appeal, the Commission was not obliged to undertake a counterfactual analysis to examine the causal relationship between the conduct and its actual and potential effects. Rather, Google could have presented such an analysis in its defence. However, the Commission may use several and different pieces of evidence than the complainant for evidentiary purposes and is not obliged to choose the same type of evidence. This does not constitute a reversal of the burden of proof by the GC and does not affect the evidential value of such a counterfactual analysis.
As- efficient competitor test
Moreover, contrary to the fourth ground of appeal, the Commission was not obliged to examine whether these practices were also capable of crowding out equally efficient competitors such as Google. It does not follow from the mere fact that Article 102 TFEU does not ensure that less efficient competitors remain on the market that any finding of an infringement of that provision is subject to proof that the conduct in question is capable of excluding an as-efficient competitor. Furthermore, it would not have been possible for the Commission to determine the efficiency of Google’s competitors in the specific case and therefore the test was not relevant to the case.
Comment
With this judgement, the ECJ mainly followed the reasoning of the Court of First Instance and the Advocate General’s Opinion. Nevertheless, these proceedings cast a light on future proceedings under Art. 6 (5) DMA and Section 19a (2) sentence 1 no. 2 ARC. This is because the ECJ emphasised more than a few times that the self-preferencing practices in question can only be considered abusive after a thorough examination of all relevant circumstances of the individual case and are therefore not per se anti-competitive. However, it is precisely this approach that forms the basis of the new regulatory solutions to self-preferencing practices in the digital sector, as can be read in the legislative documents. This brings the questions to the spotlight of antitrust discussion, whether or not these regulations meet in general and individual the principle of proportionality. It therefore remains exciting to see what the already initiated and future proceedings against Google, Amazon, Facebook, Apple, Microsoft, etc. under Art. 6 (5) DMA and Sec. 19a (2) 1 No. 2 ARC will bring to light. In any case, the Google Shopping case should not be misunderstood as a precedent to any self-preferencing cases, because the special feature of this case, the combination of self-preferencing with the simultaneous downgrading of competitors, will rarely be present in other self-preferencing cases that has nothing to do with ranking (and even there). The legal analysis may therefore differ in these cases. Practice will also have to clarify whether these proceedings really have the promised speed advantage or whether the legal issues surrounding the correct compliance measure or the proportionality objection will lead to similar procedural delays in the future.
________________________
To make sure you do not miss out on regular updates from the Kluwer Competition Law Blog, please subscribe here.
