The DMA will start to apply in March 2024. The European Commission (EC) has acquired the compromise to make the process of the DMA’s future implementation, monitoring and oversight of compliance as transparent as possible. In this same vein, the first of the DMA’s stakeholders’ workshops was held on the 5th of December on the theoretical and practical background of the ban on self-preferencing under Article 6(5) of the DMA. This entry is the overview of the first workshop, which will be followed by other entries on the subsequent events the EC will hold during 2023 around the interpretation and discussion of the different provisions of the DMA.
In the words of the head of the DMA’s task force, the choice to bring forward the first workshop held by the EC on the DMA on the particular topic of self-preferencing should not be interpreted in terms of priority setting. Instead, the workshop was centred on this provision due to the fact that it is the provision that most stakeholders have consulted about the EC in the past few months. The EC provided the space and time for stakeholders to engage with each other around the key elements of the provision, by moderating the three sessions that were held (but without directly engaging with the discussion).
Self-preferencing set out in Article 6(5) of the DMA
The DMA is not competition law. That much is clear in light of the EC’s intentions and prior statements on the matter. Thus, the DMA should not be treated as such, and neither should the ban on self-preferencing under Article 6(5). The prohibition reads as follows: “the gatekeeper shall not treat more favourably, in ranking and related indexing and crawling, services and products offered by the gatekeeper itself than similar services or products of a third party. The gatekeeper shall apply transparent, fair and non-discriminatory conditions to such ranking”.
Four key concepts stem from the provision:
- The notion of ‘not treating more favourably’ which corresponds with the principle of equal treatment;
- The object of the prohibition on self-preferencing: the processes of ranking, indexing, and crawling, which must be analysed with reference to Recitals 51 and 52 of the DMA that intend to flesh out the raison d’étre of the prohibition.
- The similarity of the services or products of competitors who do not have to bear the burden of self-preferencing.
- The application of fair, as per the definition of fairness under Recital 33, and non-discriminatory conditions.
The workshop demonstrated throughout how these four elements may be analysed across the border in a myriad of manners, stemming from the compliance that is expected by the gatekeeper by March 2024 to the proposals bounced off by rivals and stakeholders.
Common ground: a prohibition on unfair and discriminatory behaviour
Starting from the theoretical background of Article 6(5) of the DMA, the different stakeholders involved agreed upon the fact that the provision must not be read in a deterministic manner, but rather from an ontological perspective. In reality, this translates into the fact that Article 6(5) does not prohibit gatekeepers’ proprietary services or products to be shown as the most relevant result in a particular environment. Instead, the ban is centred on the process which has led the gatekeeper to produce that particular result, i.e., through fair and non-discriminatory criteria, so that it is not unbiased.
Core platform services concerned
In principle, every core platform service which is listed by the EC when designating the gatekeeper under Article 3(9) of the DMA may be affected by the prohibition. Be that in the background of the functioning of online intermediation services, online search engines, online social networking services or video-sharing platform services, amongst others. However, the workshop mainly centred its attention on the ongoing discussion around the seminal case that brought self-preferencing to life: Google Shopping (see a review of the judgment in the link here).
Although different stakeholders agreed that other forms of self-preferencing could manifest in other core platform services at large, most of the discussion revolved around how to solve the (questionable) fact that Google is non-compliant with the DMA in its current design form. A brief thought was also dedicated to Amazon’s featured offers and ranking on its marketplace regarding Amazon retail products and products catered by third-party sellers, irrespective that technical solutions were not discussed in depth.
Ideas for monitoring compliance
Even though the general requirements of Article 6(5) of the DMA remain contested between consumer associations, rival competitors on different fronts of the market and gatekeepers, the monitoring of compliance with the provision was not one of them. In this regard, stakeholders agreed that monitoring the prohibition should be quite straightforward.
Notwithstanding the fact that the ban does not address outcomes, in particular, monitoring that users may receive the most relevant results when making a query on a search engine would be quite a direct manner to address compliance by design. However, the ghost of algorithmic blackout still will loom over the prohibition in terms of enforcement, insofar as it will be very difficult to establish whether ranking, crawling, and indexing criteria may be unfair or discriminatory if the gatekeeper does not agree to unveil its search engine algorithm before the Commission.
The points of contention: the scope of Article 6(5) of the DMA
Despite that some ground rules were established amongst stakeholders, the points of discussion between gatekeepers vis-à-vis the representatives of different commercial and consumer interests occupied most of the discussion, namely on the narrow or broad reading of the disposition regarding its scope.
Direct and indirect (hybrid) forms of self-preferencing
The scope of the ban on self-preferencing must be read contextually with the DMA at large. By doing so, one of the most relevant passages of the regulatory instrument that came up during the workshop was Recital 52. The Recital explicitly interprets how the anti-circumvention clauses contained in Article 13(4) and (6) should be read in the following manner: “to ensure that this obligation is effective and cannot be circumvented, it should also apply to any measure that has an equivalent effect to the differentiated or preferential treatment in ranking”.
In the particular case of self-preferencing, this clause may run counter to the ban, insofar as the prohibition does not rely on the allocation or the replacement of particular business users in the place of the gatekeeper. Instead, it ensures that the process of reaching the result is performed with fair and non-discriminatory criteria in mind. Nonetheless, Recital 52 of the DMA infers something completely different: not only must the gatekeeper comply with the obligation, but also avoid any other form of self-preferencing if it wishes to remain compliant with the regulatory instrument. This is where the broad reading of the scope of Article 6(5) of the DMA comes from and leaves the door open for speculation. During the workshop, several examples were brought into the discussion which could also constitute a form of hybrid self-preferencing, such as the display of the news feed on social media or the mailing received by the end user following a purchase in a marketplace.
One of the most androgynous concepts, i.e., the differentiation of the services between those of the gatekeeper’s core platform services vis-à-vis its competitors, constituted a hotly debated topic across the three panels of the workshop. If the rival’s service is close enough, then the ban automatically applies to the relation of that competitor with the ecosystem holder, whereas if they cannot be said to be ‘similar’, then differentiated treatment may be justified and go under the radar of the prohibition.
On one side, some stakeholders proposed a wide reading of this requirement in the sense that the similarity between the gatekeeper’s and the competitors’ service was to be made at the level of the service and considering the characteristics in common they had at this same level. On the other side, the representatives of the gatekeeper’s interests defended a position towards interpreting the similarity between services in a much more stringent manner in light of the user’s purpose and functionality when making a query. Following this line of reasoning, few services would be similar to Google, given that the search engine itself has favoured a particular architecture choice of its own to build its ecosystem which draws end users into their verticals and prominently displayed features.
Proposals for compliance stemming from the Google Shopping decision in the background
Moving forward, concrete solutions were proposed both by stakeholders competing with Google in the search engine environment as well as by the firm itself. The devil is in the details, and the form in which Google decides to comply with the DMA in this regard will determine the future litigation of its implementation before the Courts.
Before addressing the myriad of solutions proposed along the three panels of the workshop, one particular idea was made evidently clear: no query is similar to the next. Google’s search engine is used for general queries or so-called tail queries. The former is quite self-evident for the general public, but the latter may be more intricate when considering a fully compliant solution in the terms of Article 6(5) of the DMA.
Tail queries are those that end users do in specialised areas and are particularly based on real-time data and revert a higher-monetisation rate as opposed to general queries because they bring more valuable traffic to the table (see Florence Thépot on the subject here). Some of the examples that follow from tail queries are those associated with booking a flight, a night in a hotel or finding a job offer online for a particular job position, although they can get more specific than that. The search engine does not act or display the same design and choice architecture when these types of queries are performed by the end user. Instead, the stakeholders highlighted that Google has innovated in this sense by continuously adapting its tabs and verticals to each and one of the areas concerned. Thus, it is quite clear that a one-size-fits-all design and solution to comply with the ban on self-preferencing will not be possible, given that each of these queries may deal with different economic operators who monetise their activities differently. Google agreed on this point, although not in the range of options presented by the rest of the stakeholders.
Up to seven different solutions were proposed ranging from plain interference with Google’s current design when displaying results on its verticals to others that considered nuanced solutions tailored to tail queries and specific industry contexts. For instance, building up on the experience of French auctions with gas transactions, one of the stakeholders proposed to transform the setting of the search engine environment, namely through redistributive auctions working as second price auctions (just as the existing system) where a portion of the winning bid would be redistributed amongst all bidders to maintain the sustainability of the ecosystem. The common element of all of the solutions that were proposed was that they all considered that the remedies which were sook, implemented and are currently being monitored regarding the sanctioning proceedings against Google Shopping had not produced the expected outcomes. Instead, these stakeholders agreed on the fact that Google had brought it upon itself to use this opportunity to comply with the remedies in a manner that still differentiated and favoured its proprietary services before those of its competitors.
On the note of Google’s proposal, it demonstrated its willingness to address the unfairness and lack of contestability concerns under the DMA with its proposal. In its initial stage, the solution proposed by the (not-yet-designated) gatekeeper included general principles aimed at ranking competitors on an equal footing and design options for its search engine that would enable its competitors (specifically, its competitors on comparison search services) to display their own verticals across the search engine’s results.
The first workshop organised by the European Commission came to an end with three key ideas that were repeated and reinforced throughout the day:
- There is some common ground on the general concepts around the prohibition’s purpose, i.e., Article 6(5) is aimed to defend a fair and non-discriminatory process of ranking, crawling, and indexing as far as similar services are concerned.
- The ban on self-preferencing should be interpreted by the Commission across the board, in line with the spirit of Recitals 51 and 52 as well the text of Article 13(4) and (6).
- Article 6(5) of the DMA cannot be complied with in only one single manner. Instead, compliance should be tailored to each type of query performed by the user in each sector considering industry input and stakeholder engagement. A general one-size-fits-all solution will not be enough to meet the Commission’s expectations.
All in all, the compliance of the prohibition relies on the gatekeeper and we shall wait until their particular proposals are completely fleshed out to tell whether the story of the DMA will be told in terms of success or not.
 Following the presentation at Observatorio Gonzalo Jiménez-Blanco Ashurst-ICADE, 15th November 2022.