Introduction
After several years of investigation, the Commission finally delivered its decision in the case of Facebook Marketplace, where it found that Facebook had abused its dominant position, resulting in a fine of almost € 800 million for Meta (Facebook).
According to the Commission, Facebook’s abuse consisted of engaging in:
- abusive tying practices with regard to Facebook Marketplace as users (i.e. consumers) that log into the Facebook social media platform are automatically exposed to Facebook Marketplace through: the unsolicited presentation of the Marketplace tile on the Facebook interface, the use of (jewel) notification nudging and alerting consumers to offers on Marketplace and the presentation of unsolicited listing from Marketplace on the timeline of consumers – while preventing consumers for disabling such occurrences. (section 7.2 of the decision)
- the imposition of unfair terms and conditions on competing classified ads service providers that advertise their services through Facebook and/or Instagram. According to the Commission, Facebook used the ad data generated by these parties to benefit solely Facebook Marketplace, while unreasonably burdening its competitors with requirements that are not necessary for the provision of ad services to these parties. (section 7.3 of the decision).
While the theories of harm in this case are by no means revolutionary, the decision itself represents a significant step forward in updating practice to reflect the realities of the digital economy. In doing so, the decision creates a better alignment with the DMA, which has already taken a clear stance on such practices before specific case law on these matters had a chance to develop fully.
Next to updating traditional theories of harms to the realities of the digital economy, the decision addresses several important aspects of the market definition that remain a challenging aspect of the application of Article 102 TFEU, particularly in the case of digital platforms. These aspects concern the definition of separate relevant markets for different services provided on a single platform interface, the application of the SSNDQ as an alternative to the SSNIP when dealing with zero-priced services, and the (non) existence of attention markets.
Taken together, the Commission’s decision in this case creates an important precedent for the digital economy that extends far beyond Meta’s ecosystem.
Market definition and digital platforms
The challenges concerning the market definition and digital platforms entail perhaps the first of the many challenges identified with respect to the application of (EU) competition law to these actors. Generally speaking, the main challenges in this respect concern the questions of (i) how many markets need to be defined when dealing with multisided platforms and (ii) how price-centered tools such as the SSNIP test can be applied to zero-priced products or services. In addition to these specific substantive issues, a question that has remained unaddressed in practice is the notion of attention markets, which would supposedly encompass the broader context in which digital platforms compete for consumers. The Facebook decision remarkably touches on all three points, resulting in valuable developments and insights for future practice.
i. How many markets need to be defined
The question of how many markets need to be defined has often translated in academia and practice into deciding whether defining the relevant market for a platform needs to be done for the platform as a whole (‘the single market approach’) or each of the separate sides of the platform (the multi-market approach). After years of academic debate, this distinction has now been incorporated into the new Commission notice on the definition of the relevant market (paras. 94-95). What remains challenging, however, is that this decision can be taken not only with respect to each platform but also regarding each platform service, as most of these will commonly be two or multi-sided. This latter aspect complicates the process of market definition, as it requires determining whether the various services provided on a platform are for separate markets or part of a single (bundle) market. This is a question that is typically addressed in the substantive analysis of various types of leveraging abuses of dominance, such as tying, and less so at the stage of market definition. In the case of multi-sided and multi-service platforms, this, however, is an inevitable step in the process, as the Facebook decision clearly showcases.
In this context, the Facebook decision marks the second decision concerning abuse of dominance, in which different services offered on a single platform interface are considered to form part of separate relevant markets; the first case being Google Shopping. In the case of Facebook, the Commission found that the Marketplace service, the social media service, and the (display) advertisement service offered on the Facebook interface were part of separate relevant markets (para. 104). The definition of markets in the decision offers a combination of the single market and multi-market approaches. The multi-market approach can be seen with regard to the social media aspect of Facebook, where the user side of the service forms part of a different market than the service that monetizes it, namely the (display) advertising service offered to publishers. The single market approach is taken with regard to the Marketplace service, for which one overall market is defined, consisting solely of the market for online classified ads services (OCAS). In doing so, the decision establishes an important precedent for defining relevant markets when dealing with multi-sided platforms, clarifying that the single- and multi-market approaches are not mutually exclusive and may need to be adopted in conjunction with regard to a single multi-sided and multi-service platform.
ii. The SSNIP test in zero-priced settings
The difficulties associated with applying the SSNIP test in zero-priced settings were identified early on in the discourse on the application of competition law to digital platforms, which often offer some of their services at no charge. After extensive debates, the solution was found in adjusting the price-centered SSNIP test into a quality-centered SSNDQ test, where a theoretical decrease in quality is assessed as an alternative to the theoretical increase in price of the SSNIP test. This solution was first implemented in Google Android (paras. 286-300), and later officially included in the new Commission notice on the definition of the relevant market (paras. 96-98). Although this choice is welcome in practice, a related matter that remains unanswered is the scope of the theorized decrease in quality; i.e., should it be 5-10%, as with the SSNIP, or more. When presented with the opportunity to provide further clarity, the General Court stated that the Commission is not obligated to set and communicate a specific figure for this purpose (para. 180). While this is understandable, as quality and reactions to its decrease may vary across sectors, the absence of guidance in this regard inevitably gives rise to potential conflicts, as seen in the case of Google Android and now with Facebook. Here, once more, the degree of theoretical decrease in quality used for the SSNDQ by the Commission was challenged (paras. 386-407).
This issue does not concern the quality of the decision per se, but rather one that highlights a missed opportunity in the context of the new notice on market definition (see here for an extensive discussion on this point). It is evident that a set range of figures will not work well in the context of the SSNDQ since quality, unlike price, is a multifaceted concept. Nevertheless, there is room for improvement when it comes to the process that determines how the SSNDQ is operationalized in general terms, so as to give undertakings a better idea of what that test could look like when applied to their products or services. This would not only help prevent further challenges in the context of Article 102 TFEU cases but potentially also those falling under Article 101 TFEU as well as under the EUMR, where market definition plays an (almost) equally important role (see discussion also here).
On-platform tying – A new form of tying practices in the context of multisided platforms
The type of tying addressed by the Commission in the Facebook case opens the door to identifying a new form of anti-competitive tying in the context of multisided platforms, specifically on-platform tying. In such a context, on-platform tying would entail situations where the various services facilitated by a multisided platform are tied to each other. In practice, this would entail the use and/or participation of one service (i) is made conditional upon the use of or participation in another service; or (ii) that the use of one service automatically triggers the use of or participation in another platform service. This kind of practice has not yet received much attention in the context of competition policy. Most of the focus in this regard has been on cross-platform tying, where two or more separate platforms are tied (e.g., Google Android). From an enforcement perspective, this is unfortunate because the two types of tying can produce similar if not identical anti-competitive effects.
Generally speaking, the anti-competitive potential of tying practices has been considered to result in (i) foreclosure in the tying and/or tied product markets; (ii) deterrence of entrance in the tying and/or tied product markets as well as a (theoretical) third product market for a novel product capable of replacing the combination of the tying and tied product; (iv) the extraction of supra-competitive prices in both tying and/or tied product markets. The manifestation of such potential harms in practice depends, however, on the circumstances of each case and the market conditions present at the time of the analysis. In the context of multisided platforms, such as Facebook, similar findings have been reported (see more extensive discussions here and here). The fact that the tying and/or tied products or services are provided for free (zero-priced) does not detract from the potential anti-competitive concerns. Quite the contrary, it is precisely these circumstances of zero pricing that increase the likelihood of tying practices. If platforms are unable to compete on prices for their goods or services, they must find a way to enhance their zero-priced offer to consumers or commercial trading parties while competing against their rivals. In this regard, it is good that the Commission categorically dismissed Facebook’s argument on this point (paras. 807-808).
Both on-platform and cross-platform tying allow the dominant undertaking to leverage part of its customer base onto a new platform or platform service without having to face the so-called chicken-and-egg coordination problem. This makes such practices very effective for ‘envelopment attacks’ for defensive or offensive leveraging purposes. Such practices will typically be applied with respect to the end-consumer side of multisided platforms because, for the leveraging to work, the leveraged customers need to be users of both the tying and tied platforms (in case of cross-platform tying) or platform services (in case of on-platform tying). Typically, this overlap tends to occur only on the end-consumer side of multi-sided platforms. The potential of such a practice to facilitate significant market power leveraging across markets will depend on the degree of overlap between the two platforms or the platforms’ services with respect to these end consumers. The greater the degree of overlap, the greater the potential for leveraging. In the case of Facebook, although the two services (social media and marketplace services) are commercially unrelated, they share a tremendous degree of (potential) overlap when it comes to end consumers (see extensive discussion here).
Against this background, the Commission’s case against Facebook creates an important precedent for the enforcement of Article 102 TFEU vis-à-vis digital platforms. In the long run, all platforms reach a point where they will need or wish to expand for strategic and/or financial reasons. At this point, tying practices will offer an attractive and effective strategy, as any successful expansion entails, in essence, a market power leveraging exercise. Although the addition of new services to a platform interface is not inherently harmful, forcefully securing the joint use of such services is. The Facebook Marketplace case clarifies that coercive strategies aimed at leveraging the platform across services cannot escape scrutiny simply because such services are offered on one interface. In this regard, the case also aligns more closely with the DMA, which addresses similar practices under Articles 5 and 6.
The positive side of the effects-based approach
The effects-based approach, also known as the more economic approach, under Article 102 TFEU has been a divisive issue over the past few years. While some welcome this development for the economic rigor it brings, most view it as a complicating factor in enforcing this provision. The more evidence the Commission (or NCA) is expected to bring to substantiate a claim, the more challenging a case is perceived to be. To what extent the standard of proof in Article 102 TFEU is genuinely intensified by this approach remains an open discussion. While there is no doubt that a purely formalistic approach to abuses is insufficient, the recent case law, including Android Auto and the Opinion of AG Kokott in Google Android, indicates that finding an abuse primarily requires establishing the capability to lead to an exclusionary effect and/or departing from competition on the merits. Proving such circumstances to the requisite standard should arguably not be too complex, certainly when compared, for example, with establishing dominance.
Assuming, however, that the effects-based approach truly requires engaging in more rigorous economic analysis, the case of Facebook Marketplace demonstrates why this may be very advantageous for enforcement in some cases. The prohibited conduct in Facebook Marketplace could, on the face of it, easily be mistaken for legitimate practice, as correctly understanding the intricacies of these practices requires looking beneath the surface. Were it not for the effects-based approach that prompted the Commission to conduct an intensive assessment of Facebook’s entire architecture and its impact on consumer decision-making, it is hard to see how this abuse could have been identified. In this regard, the evidence of special notifications, signaling, and presentation that Facebook linked to the Marketplace service was decisive for qualifying the practices as abusive tying and demonstrating the anti-competitive capability.
Just as in Google Shopping, abuse is not a matter of a single action, namely the inclusion of the Marketplace tile in the Facebook interface, but rather a collection of practices that accompany it in ways that deviate from competition on the merits. Unlike in Google Shopping, however, Facebook’s conduct, when viewed in this context, fits neatly within the framework of abusive tying. In this sense, the theory of harm and the form of abuse in Facebook Market are neither novel nor controversial. Nevertheless, the Facebook Marketplace case is of great importance for future practice in the digital economy, where the notion of coercion can take different forms, including (hyper) nudging and the creation of so-called dark patterns. Such practices may appear to allow for consumer choice, but, in practice, entail a harmful choice architecture that aims to achieve precisely the opposite by tapping into the cognitive and emotional weaknesses of consumers.
Against this background, the Facebook Marketplace case demonstrates that while the effects-based approach may complicate the enforcement of evident cases, it can be a valuable tool for uncovering complex and covert anticompetitive conduct, thereby advancing the enforcement capability of current practice.
Clarifying the framework of unfair trading conditions
In addition to modernizing the frameworks for tying to the realities of the digital economy, the Facebook Marketplace decision brings more clarity with regard to the legal test of unfair trading conditions. For some time now, the legal test for this abuse has taken various forms and constellations in the Commission’s case practice and the EU courts’ case law. While the legal test criteria applied consistently appeared across cases, they were not cumulatively applied in a single case. This resulted in a lack of clarity as it was not evident to what extent the criteria mentioned in previous cases were indeed (strictly) cumulative. Recently, in the case of the Apple App Store (Music Streaming), the Commission addressed this matter by specifically indicating how the legal test for this abuse is to take shape (para. 529). According to the Commission this the test for finding an abuse based on unfair trading conditions requires establishing that the dominant undertaking (i) imposed trading conditions on its trading partners that are (ii) unfavourable or detrimental to their interests of those of third parties, and (iii) are not necessary for the achievement of a legitimate objective, or in any event not proportionate for that purpose in that they go beyond what is strictly necessary.
Now, with the Facebook Marketplace decision, the Commission confirms once more that this is the course it will take in the future when enforcing this type of abuse (para. 1071). Here, in its application to Facebook’s data collection and accumulation strategies, the Commission’s application of the test for unfair trading conditions offers significant added value for the case itself as well as for future practice. For the case itself, it offers a structured approach to the analysis of Facebook’s actions that ensures legal certainty. Here too,
Beyond the ambit of this case, the strengthening of the legal framework for unfair trading conditions abuses is one of the main, if not most important, contributions of the decision for future practice. Where, in the past, unfair trading conditions abuses were treated as a residual category of abuse under Article 102 TFEU, that will not be the case in the context of the digital economy. In this complex and constantly evolving context, the implementation of unfair trading conditions that do not fall neatly within the frameworks of other abuses is likely to be the rule rather than the exception. This is due to two main ingredients that cases against dominant multisided platforms involve: (i) a significant imbalance of market power between the platform and its competitors, trading parties, and, of course, consumers; and (ii) complex business models and services that spread across markets implemented through practices characterized by both exploitative and exclusionary traits.
By introducing clarity and structure to the framework for unfair trading abuses, the Commission (re)gains access to another device in its legal toolkit that has been (unjustly) underused so far. In doing so, the Commission significantly reduces the need for continuous exploration of the open-ended boundaries of Article 102 TFEU, and thereby also the risk of becoming entangled in more prolonged sagas akin to Google Shopping. While novel cases may still require a tailored approach under Article 102 TFEU, a large part, if not most, of the cases could fit well within the ambit of unfair trading conditions. In this regard, the use of unfair trading conditions as a ground for establishing abuse, if done based on the Commission’s approach, is far less likely to be successfully challenged based on legal certainty arguments, as legal certainty is precisely what it delivers.
Restorative remedies
Over the years, the focus of competition law enforcement has traditionally been on identifying infringements. This has been particularly true of enforcement in the digital economy over the past decade. This focus is to some extent justified by the fact that the ability of current practice to address anti-competitive practice in this context has regularly been challenged, and is often found lacking, resulting in the implementation of the DMA. Nevertheless, this focus negates the fact that effective enforcement is not only about identifying anti-competitive conduct but also remedying it. There is little comfort in finding infringements if these cannot be brought to an end. In practice, however, this key aspect of enforcement is often overlooked. The prevailing assumption is that cease and desist orders, or other measures that mirror the infringing practices (e.g., untying products or services, or providing equal treatment instead of discriminatory treatment) will be sufficient, which is hardly the case.
Generally speaking, the primary objectives of competition law remedies are to terminate the infringement, prevent recurrence, and restore or re-establish the state of competition. The current approach to remedies has been mostly limited to the first two objectives, under the assumption that the market will handle the rest. It is not hard to see why this approach is flawed. As much faith as competition law enthusiasts may have in the market, the truth remains that markets do not self-correct in all circumstances. Accordingly, in some cases, specifically designed restorative measures may be required to ensure effective enforcement. This can be seen in the case of Akzo, where the Commission indicated that Akzo was prohibited from offering its rivals’ clients lower prices than it offered its own customers (see article 3 of the decision). This prohibition applied to the customers that Akzo unlawfully took from its competitors as well as the existing customers of its rivals. These measures, confirmed by the CJEU (see paras. 155-157), allowed Akzo’s rivals to regain the clients they had lost to Akzo due to predatory practices and restore the state of competition that existed before Akzo implemented its abusive practices.
Without such measures, the anti-competitive effects caused by the infringement will either stay in place or, worse, continue to amplify even after the anti-competitive conduct has been abandoned. This latter scenario precisely describes the risk posed by infringements involving multisided digital platforms due to the network effect at play in such cases.
In cases where platforms adopt anti-competitive practices that amplify the network effects associated with their business model, the self-reinforcing loop caused by such a process cannot be halted simply by requiring the undertaking concerned to terminate its conduct. This is because the abandonment of anti-competitive conduct will not likely change anything in the value the platform is perceived to provide its customers. For example, Apple managed to impose Apple Pay on consumers and trading parties through various anti-competitive practices related to the App Store, as well as access to its NFC chip. These actions have arguably fuelled the popularity of both the App Store and Apple Pay for years. Such practices created clear biases and consumers’ habits to the detriment of (potential) competitors (and consumers themselves), which will not seamlessly change now that the Commission actively pursues such aspects. If most shops and app developers continue to rely on Apple Pay, there’s little to prompt consumers to reconsider their use of it, even if the initial adoption was a result of anti-competitive conduct. The same can also be seen in the cases of the Android Play Store and Google Chrome; the mere untying of these products will not necessarily cause consumers to turn to them less often after years of habit caused by anticompetitive conduct.
Against this background, the Commission’s approach to remedies in Facebook Marketplace is a welcome development that is nothing short of a necessity when dealing with digital platforms. In essence, the remedies options proposed by the Commission (para. 1404) aim to tackle the bias triggered by its anticompetitive tying practices. Accordingly, Facebook can choose between: (i) separating the Facebook Marketplace from the social media platform and offering it on a standalone basis; or (ii) creating something akin to a choice screen that allows consumers to choose neutrally one or several OCAS providers which they wish to have embedded into Facebook for their usage.
Such measures allow creating a moment where consumers are asked to actively (re)consider their choices that have been previously influenced by such abusive practices, without which the effects thereof would likely persist. Here, it is also worth noting that in addition to these two alternatives, the Commission presented Facebook with the possibility for Facebook to come up with its own design. By doing so, the Commission pre-empted any complaints from Facebook concerning proportionality or legal certainty, as it offered all the leeway one could ask for.
The restorative aspect of remedies is also pursued with respect to the unfair trading conditions affecting OCAS providers regarding ad-related data (para. 1406). Here, the Commission required Facebook to ensure both legally and technically that non-publicly available ads-related data derived from or provided by OCAS providers when they purchase Meta’s online advertising services cannot be used by Meta either within Facebook Marketplace or within Facebook to the advantage of Facebook Marketplace. This is reminiscent of the Amazon Marketplace commitments decision and the obligations of Gatekeepers under Article 6(2) of the DMA. Importantly, however, the Commission adds a very important component to this remedy that will be key to many future cases. According to the Commission, to the extent that Facebook already benefited from the use of the ads-related data, any such effects would have to be removed, in particular, via a retraining of any relevant machine learning Marketplace models. In other words, the data-related advantages gained by Facebook through its anticompetitive conduct must be neutralized.
The importance of this latter point cannot be overstated. In the digital economy, where dominant actors have market valuations running in the trillions, competition law fines can be seen by them as the cost of doing business. One cannot speak of deterrence in cases where multibillion-euro fines can be recovered in a matter of months. Deterrence is even weaker if the undertaking concerned is allowed to retain its unfairly gained advantage. By requiring Facebook to train its algorithms, the Commission aims to eliminate the incentive to continue infringing competition law by taking away the benefit of such infringement and making it far more costly to engage in. This approach is not only key for the Facebook Marketplace, but it will also likely become increasingly used in future cases, as access to data and the insights it provides will often be a primary motive behind anticompetitive conduct in the digital economy.
Outlook
The Facebook Marketplace will undoubtedly become one of the leading cases for the application of Article 102 TFEU to multi-sided digital platforms. The decision effectively addresses multiple matters that required clarification in this challenging context. The decision clearly demonstrates that the Commission is well aware of these issues and offers valuable guidance for future cases. In this regard, the remaining difficulties with the SSNDQ are likely to be a calculated compromise rather than an omission. As expected, Meta appealed the decision, and it is now up to the GC to determine the course practice will be heading. For now, the decision marks yet another significant step in updating current practice, which has fortunately not slowed down after the implementation of the DMA.