Until recent times, an interim measure decision of the Turkish Competition Authority (“TCA”) was something hard to come by. However, with the growing prominence of digital markets and the equally strong response from the competition authorities, that situation has changed as well. The latest experience shows that the TCA is more determined to actively use this tool in an attempt to efficiently respond to conducts in these fast-evolving markets.
Among those decisions, the Trendyol Interim Measures Decision (or Trendyol Decision) clearly was the one that attracted the most interest and debate. So it was very exciting to get informed that on 25.05.2022, the administrative court has reversed in part and affirmed in part this decision via a detailed analysis of each interim measure (“Court’s Decision”).
As mentioned, the ongoing investigation case on Trendyol, an e-marketplace owned by the Alibaba, and the Trendyol Decision attracted a lot of interest both from practitioners and from market players since they are the first sightings of the shadows casted over by the rumoured DMA-like legislations. Back in September last year, via its announcement on Trendyol Decision, the Authority has also made clear that the underlying investigation is about whether Trendyol infringed the Competition Law by favouring its brands over the ones sold by other sellers through its website, by using the data that it obtained through its marketplace operations in favour of its own products and by discriminating among the sellers using the platform.
Based on its provisional findings, in an attempt to limit the effects of these conducts, the TCA has adopted the Trendyol Decision, a decision comprised of three interim measures to counter the three identified conducts and of other four measures to ensure effective compliance. In its own words the TCA announced the interim measures as: “In this context, it has been decided that Trendyol refrains from all behaviours aimed at favouring its own products, discriminating between sellers, sharing and using data obtained from marketplace activities in favour of its own brands, including interventions made through algorithms and coding, and should take all necessary technical, administrative and organizational measures to ensure the auditability of these measures”.
Since the aim of this article is to analyse the Court’s very extensive review of the interim measures brought by the Trendyol Decision within the scope of the Article 9/4, it is beneficial to recall the exact scope of the interim measures, according to which Trendyol shall;
- within the scope of its marketplace activity, put an end to all kinds of conducts, behaviours and practices, regarding products and services offered under its own economic entity, including interventions made through algorithms and coding, which provides an advantage against its competitors, and avoid these conducts in the continuation of the investigation process.
- stop the sharing and use of all kinds of data obtained and produced from its marketplace activity, for products and services offered under its own economic entity, and avoid these conducts in the continuation of the investigation process,
- put an end to all kinds of conducts, behaviours and practices, including interventions made through algorithms and coding, which may discriminate among sellers operating in the marketplace, and avoid these behaviours in the continuation of the investigation process,
- take all necessary technical, administrative and organizational measures to ensure the auditability of the above-mentioned interim measures,
- keep the parametric and structural changes made on all algorithm models used for the purposes of product search, seller listing, seller score calculation, etc. at least 8 (eight) years, with old versions and undeniable accuracy,
- keep the source codes of all software specially developed for use within the company with old versions and undeniable accuracy, for at least 8 (eight) years,
- keep user access and authorization records and manager audit records for all software used within the scope of the execution of business processes within the company for at least 8 (eight) years, with undeniable accuracy.
Against this list of interim measures, Trendyol took the case to the first-degree administrative court. According to the Court’s Decision, Trendyol in its essence argued that:
- the obligations listed under the umbrella of interim measures are not in compliance with the rules of Competition Law,
- the possibility of serious and irreparable damages until the final decision is made cannot be demonstrated [by the Trendyol Decision] concretely with legally valid information and documents,
- the decision being established on abstract justification is not in line with the principle of the rule of law,
- the evidence collected by the unauthorized authorities and illegally, cannot be a basis for the decision of interim measures, and
- a significant part of the documents belongs to a period where the companies were not in a dominant position in the market.
As a first step of its reasoning, the Court puts forward the three conducts identified by the TCA as the basis for the first three interim measures. According to the TCA’s own documents these are:
- Intervening in the algorithm in a way that would provide an advantage to the products it offered for sale in its role as a retailer and providing the next day delivery opportunity only for its own products,
- Using the data of the sellers active in the marketplace in the preparation of the marketing/design strategy that will provide an advantage to its own retail activity, and
- Discriminating between vendors selling in the marketplace, through interventions in the algorithm and the lack of transparency for sponsored products.
In its decision the Court holds a two-step analysis for evaluating the interim measures. First, by examining the available evidence in the case file, the Court looks whether the conditions in Article 9/4 are satisfied. As a reminder, pursuant to Article 9/4 of the Competition Law, “where serious and irreparable harm is likely to occur before the final decision, the Board may take interim measures in order to maintain the situation before the infringement, without exceeding the scope of the final decision”. In the second stage, the Court then analysis whether the drafted interim measures properly address the relevant concerns. Also at this point the Court notes that the ”infringement” in the text of the article refers to a “plausible existence of an infringement” since if the adoption of the former interpretation would not be in line with the reasoning of the Competition Law.
As mentioned, first, the Court goes on to analyse each of the evidence that is used to justify the first three interim measures aimed individually at each of the investigated conducts. The decision shows that while doing so the Court had requested additional information about the documents from Trendyol, presumably to be able to assess fully the background of the conducts. As will be examined in detail, in its analysis the Court uses this and other information:
- to check the validity of the evidence to see if they support a strong plausible existence of a violation, an “abuse” of dominance, that will justify the application of interim measures,
- to check whether serious and irreparable harm is likely to occur before the final decision, mostly blended together with the first condition, and
- to check whether the conduct can be objectively justified.
First Interim Measure
Regarding the first interim measure targeting the first conduct, the Court examines 14 documents, which are mostly screenshots of work orders made to software engineers that look to be requesting certain changes in the algorithm in favour of platform’s own brands.
For the first seven documents, the Court finds that these belong to year 2017, a period for which we understand that there is not a dominance finding made by the TCA against Trendyol. Depending on this finding, the court decides that these documents cannot be used as a basis for an interim measure within the scope of the current investigation. At this point, it should be pointed out that the Court’s Decision does not check whether the conducts identified in these documents also carried over to the periods under investigation where the TCA finds Trendyol as dominant.
For Document-8 which the TCA argues as showing Trendyol removing “seller-scores” for its own activities and thus favouring itself against other sellers. However, the Court sides with Trendyol which brings the explanation that since Trendyol has multiple ways of selling products (its own products, reselling products of others and through its warehouses) a single score would not be feasible. The Court uses this explanation to argue that Trendyol and the sellers are not in “equal status” and thus the conduct cannot be categorized as an abuse. This part of the decision seems to argue that for self-preferencing to be considered as an abuse, the dominant undertaking and the sellers should be assessed to see whether they are on equal status just like it would be in a downstream discrimination case.
Similarly for Document-9, the Court, in essence, using the information supplied by Trendyol, concludes that the company did not favour its own operations while offering delivery services to the sellers.
Lastly, the Court examines Documents 10-13 which are argued by the TCA as showing that Trendyol has altered the search algorithm to make sure that its brands come on top in searches. Against the argument brought by the Trendyol which asserts that the conduct has been put in place because of the complaints received from consumers complaining about not finding Trendyol brands, the Court requests these complaints to be submitted to the file. This approach of the Court suggests that the Court is open to let Trendyol justify the adjustments made to the algorithm. However, since three complaints submitted by Trendyol, belong to dates after the conduct, the Court concludes that the Documents 10-13 “give rise to strong suspicion that [Trendyol] interferes with the algorithm in a way that will provide an advantage to the products offered for sale as a retailer by not including the named brands in the normal sorting filter and by carrying them to the higher ranks with computer coding”. Similarly for Document-14, which shows that Trendyol removed the follower count of its own brands, the Court concludes that “It is strongly suspected that Trendyol interferes with the algorithm in a way that will take advantage of the products offered for sale in its role as a retailer, by not undermining the credibility of Trendyol brands with few followers and by favouring its own products.”.
Accordingly, the Court affirms the fulfilment of Article 9/4 conditions for the first conduct based on these final documents.
Second Interim Measure
As for the second conduct, by surveying the Documents 15 through 20, the Court observes that the documents under consideration are correspondences between Trendyol employees which show that the sales data of competing brands are accessed and used within the company via tables and analysis tools. Although Trendyol has argued that the data is accessible by other sellers, the Court rejected this argument given the lack of proof. Therefore, the Court has concluded that the required conditions by the Article 9/4 are also fulfilled for the second conduct.
Third Interim Measure
Finally, and surely most interestingly since the Court annuls the interim measure brought against this third category of conduct, the Court examines the Documents 21 and 22. According to the Trendyol Decision, these two documents, which are intra-company correspondences, show that Trendyol can alter the search results to rank some sellers higher. The Court inquired about the documents, and from the response it received, concluded that the correspondences are made just to identify “errors” in the system and the sellers are not ranked higher than they should be.
However, after this initial point, the analysis of the Court makes a significant shift from its analysis under the first two conducts. As shown, under its examination of the first two conducts, the Court is more interested in whether the evidence supports the conducts put forward by the TCA rather than whether the conducts in themselves could be regarded as an abuse of dominance. This is interesting since the abuses identified by the first two conducts are very novel in nature not just for Turkey but for the world. Accordingly, for the third conduct as well, it is expected that the Court just concludes on the lack of existence of the conduct to annul the third interim measure.
However, for the discrimination allegations, the Court goes onto observe that the TCA did not analyse whether the sellers are of equal status or whether this situation is serious enough to amount to an infringement of the Competition Law. Furthermore, the Court argues that it would be pro-competitive to treat new sellers more favourably since the support given to new sellers would increase the competition in the market. It is highly surprising to witness that while on one hand the Court does not go into detail to analyse more novel conducts such as self-preferencing and use of sellers’ data to see whether they amount to abuses of dominance and while on the other argues that the TCA should have made a more detailed analysis, about a more familiar conduct such as discrimination.
Lastly and more importantly, the Court makes the following observation:
“It is not legally possible to apply a measure by making an evaluation in the form of a final decision, it is only possible with a final decision to decide whether the plaintiff discriminates between the sellers or not, and if discriminates, whether this situation is an abuse of dominant position can only be revealed with the final decision, it is clear that it cannot be determined that the dominant position is abused by the plaintiff via discriminating between the sellers and that a situation on the contrary will result in the use of the interim measure as the main punishment tool by exceeding the purpose of the interim measure, and this situation will not be compatible with the principle of the rule of law.”
It is very hard to interpret from the Court’s Decision exactly how the approach of TCA changes between conducts 3 and 1-2 so that the Court opposes to the third interim measure on the grounds that if the TCA’s approach is accepted it would amount to a determination of violation and thus a replacement of the final decision. In all three conducts, the TCA puts forward the existence of a conduct that would possibly be considered as an abuse of dominance. In the first two, the Court only checks the evidence to see whether they support the existence of the conduct while on the third one it goes on to argue that ordering an interim measure would mean a conclusion on the existence of a violation. Accordingly, the Court’s approach that it brought to the third conduct seems disjointed from the rest of its decision.
As a result of these assessments, the Court upholds the first two interim measures while annulling the third one. In the second part, the Court examines whether the interim measures brought are relevant to the identified conducts. However, since the interim measures are drafted in a way that literally request stopping the alleged “conducts”, the Court affirms the two measures corresponding to the first two conducts.
Regarding the measures brought in for effective compliance, the Court unsurprisingly annuls the periods set for measures 5-8 which extend beyond the final decision for a total period of 8 years. Since the objective of the interim measures is to preserve the competitive conditions from getting worse, there is no doubt that they should last until the final decision where the TCA will have the discretion to order the measures it seem necessary, anyway.
As a final comment, we believe that the decision of the Court is very important for various reasons. First, it shows that the Courts are ready to make an effective evaluation of the interim measure decisions which are powerful tools that should be carefully utilized. Second, while doing so the Court does not refrain from going into details of the case and from requesting further information in that regard. And finally, this decision makes it clear that the interim measures should not be extended to periods going beyond the final decision.