The Turkish Competition Board’s Amazon Decision: Algorithmic Pricing Tools and the Hub-and-Spoke Coordination Allegation

1. Background of Investigation

The process began in August 2023 with a preliminary investigation concerning Hepsiburada. At the conclusion of that preliminary investigation, it was determined that no full investigation needed to be opened with respect to the most-favored-nation (MFN) clause in Hepsiburada’s contracts and its discriminatory practices; however, in October 2023, a decision was made to open a full investigation to determine whether the automated pricing mechanism infringed Law No. 4054. Having established that Trendyol and Amazon employed similar mechanisms, the Board simultaneously launched ex officio full investigations into those undertakings as well. The investigation period was extended by six months by a decision dated March 21, 2024.

During the investigation process, Trendyol, Hepsiburada, and Amazon each requested to submit commitments. Having found the commitments offered by Trendyol and Hepsiburada sufficient to address the competitive concerns, the Board closed the investigations against those two undertakings in October 2024.[3] Amazon, for its part, withdrew from the commitment process, arguing that, contrary to the allegations, the use of the automated pricing mechanism had not given rise to any anticompetitive concerns. The Board accordingly terminated Amazon’s commitment process on October 10, 2024, and decided to continue the investigation.

2. Relevant Market and Amazon’s Market Power

Amazon has been operating in Türkiye as a multi-category marketplace platform since September 2018. In addition to its e-marketplace service, the company has offered the Amazon Prime subscription program since September 2020; Prime Video is also available in Türkiye as part of the Prime subscription. Amazon is furthermore the owner of Twitch Interactive, Inc.

For purposes of the relevant product market definition, the substitutability analysis examined the relationships among (i) online and physical channels, (ii) e-marketplaces versus sellers’ own websites and social media channels, and (iii) multi-category and single-category e-marketplaces. Drawing on sector reports, prior Board decisions,[4] and the information and documents in the case file, the Board concluded that online channels do not constitute a substitute for physical channels, that e-marketplaces do not constitute a substitute for sellers’ own channels, and that multi-category marketplaces do not constitute a substitute for single-category marketplaces; accordingly, the Board defined the relevant product market as the “multi-category e-marketplace market.” The geographic market was defined as Türkiye.

In the market power analysis, the Board noted that, owing to the multi-sided structure of the relevant market, traditional market-share metrics alone are insufficient; user numbers, network effects, barriers to entry, and other structural indicators must also be taken into account alongside transaction volume and transaction counts. In the case file, market size and the competitive position of the undertakings were analyzed primarily on the basis of transaction volume and transaction counts, with total transaction volume intermediated by each platform serving as the principal benchmark. The analysis found that Amazon’s share measured by transaction volume had increased in the period following its market entry, then declined partially before stabilizing in third place, while its share measured by transaction counts peaked during a certain period before falling to fourth place. Taken together, the transaction volume and transaction count figures led the Board to conclude that Amazon occupies the position of the third-strongest player in the multi-category e-marketplace market in Türkiye.

3. Amazon’s Automated Pricing Mechanism

Amazon introduced its automated pricing mechanism in Türkiye as of April 2020. The mechanism is designed to automate Buy Box (Competitive Purchase Box) competition in cases where the same barcoded product is offered by multiple sellers. The “Competitive Featured Offer” displayed in the Buy Box is the offer that, according to the platform’s algorithmic metrics, is deemed to deliver the highest value to the customer; “Add to Cart” and “Buy Now” transactions are completed through this offer. Visibility in the Buy Box is therefore of critical importance to sellers.

While not mandatory, sellers may define automated pricing rules and automate price changes within specified parameters through the Seller Central interface. The system offers sellers four types of rules: “Competitive Featured Offer,” “Competitive Lowest Price,” “External Competitor’s Price,” and “Unit Sales-Based.” The first three rules are price-referenced and give sellers the option of matching or undercutting the reference price. The “Unit Sales-Based” rule triggers price changes based on sales performance.

Amazon stated that the mechanism is optional, that sellers determine their prices independently, and that the Featured Offer is not based solely on price but also on multiple parameters such as inventory, delivery performance, and seller reliability. Amazon also argued that the proportion of sellers using automated pricing represents a relatively small share of total sellers.

The Board noted that the Competitive Featured Offer is determined through a dynamic, multi-parametric, and algorithmic process in which price is not the sole determinant and in which factors such as inventory status, delivery performance, and seller reliability are also taken into account. The Board further stated that the technical details concerning the operation of this algorithm are not disclosed to sellers; sellers are only informed about general performance criteria.

4. Anticompetitive Concerns Raised by Pricing Algorithms

The Board considered it necessary to identify the potential effects of pricing algorithms on competition as part of the case file. It noted that, in tandem with digitalization, undertakings are making widespread use of algorithms for pricing, personalization, and market forecasting purposes, and that such tools may be classified as data monitoring, dynamic pricing, and personalization algorithms. The Board acknowledged that algorithms can produce procompetitive efficiency gains, including cost reductions, dynamic pricing capabilities, and lower consumer search costs.

At the same time, the increased market transparency and the capacity for rapid responses to price changes that algorithms afford may give rise to the risk of tacit coordination among undertakings. The Board emphasized that, particularly with respect to pricing algorithms, the possibility of parallel conduct becoming stable in the absence of any formal agreement must be evaluated from a competition law perspective.

In the academic literature, the effects of algorithms on competition are examined under the headings of coordinated behavior risks and unilateral behavior risks. It is noted that algorithms may be used as instruments for committing infringements such as price-fixing, information exchange, or bid-rigging, and that a hub-and-spoke structure in particular may emerge through a third-party software provider. The Board stated that certain cumulative conditions must be satisfied for a structure analogous to the hub-and-spoke model to be assessable under Article 4 of Law No. 4054. Specifically:

1. A seller must communicate its forward-looking pricing intention to the platform;
2. The platform must transmit that information to other sellers;
3. The seller receiving the information must be in a position to foresee the originating seller’s pricing strategy; and
4. That information must be knowingly used in formulating the recipient seller’s pricing strategy.

The Board stated that algorithms do not produce anticompetitive outcomes in every case and that the legal assessment must be conducted on a case-by-case basis, taking into account such factors as the design and intended use of the algorithm, the contractual relationships between the parties, and the foreseeability of any potential anticompetitive outcomes.

The case file noted that no prior Board decision of a comparable nature existed, but that the decisional practice of foreign authorities is instructive for understanding the circumstances under which algorithms may produce anticompetitive effects. In this regard, the Board cited the Court of Justice of the European Union’s (CJEU) judgment in Eturas,[5] the Danish Competition and Consumer Authority’s (DCCA) Ageras decision,[6] the Competition and Markets Authority’s (CMA) Trod–GBE decision,[7] and the RealPage[8] and Uber[9] cases litigated in the United States.

5. The Board’s Assessments

During the preliminary inquiry phase, an opinion was obtained from the Department of Information Technologies regarding the automated pricing mechanism. Although Hepsiburada was the inquired undertaking at the time the opinion was issued, it was noted that the automated pricing mechanisms used by both platforms share a structural similarity in that both are rule-based. The Department of Information Technologies stated that through automated pricing, sellers are able to set their prices below, above, or equal to the winning Buy Box offer, and that this structure may increase the likelihood of coordination among sellers and may facilitate price alignment through the algorithm.

During the full investigation phase, an analysis was requested from the Department of Economic Analysis and Research on whether the automated pricing mechanisms produced coordinated effects. The Department of Economic Analysis and Research noted that the systems under examination are rule-based, contain no machine-learning component, and allow sellers to match or undercut specified reference prices by command. Accordingly, the potential competitive risks were assessed as being more limited than those posed by learning-based systems. The relatively low proportion of sellers using automated pricing was also noted.

Hourly data from 67 Amazon sellers across 7 different products spanning the period January 1, 2024, through May 31, 2024, were examined. The analysis found that price movements were heterogeneous and dynamic, and that no positive finding of coordination or anticompetitive pricing behavior could be reached. DEAR cautioned, however, that the results should be interpreted with circumspection given the low usage rate and lack of continuity, and specifically flagged the risk that widespread adoption of the “match reference price” rule could reduce price diversity.

Structuring its analysis under Article 4 of Law No. 4054, the Board assessed, in turn, the legal characterization of the automated pricing mechanism, the potential competitive risk scenarios, and the reasons why an infringement finding was not reached in the specific case. In this context, the Board recalled that Article 4 covers both horizontal and vertical agreements, and stated that it was first necessary to determine the nature of the relationship between Amazon and its sellers. Drawing on the definition set out in Communiqué No. 2002/2 on Block Exemption for Vertical Agreements, the Board accepted that the sellers are undertakings offering products on the marketplace and that Amazon is a provider of platform services connecting sellers with consumers; it accordingly concluded that the parties operate at different levels of the production and distribution chain. Finding that the automated pricing mechanism forms part of this relationship, the Board determined that the relationship is vertical in nature within the meaning of Communiqué No. 2002/2.

The Board emphasized that, irrespective of whether the relationship is characterized as horizontal or vertical, the determinative criterion under Article 4 is “object” and/or “effect”; to establish an infringement, it is necessary to demonstrate the existence of an agreement or concerted practice having the object or effect of preventing, distorting, or restricting competition. In this connection, the Board recalled that the fundamental objective of competition law is to ensure that each undertaking determines its commercial strategy independently of its competitors.

While the Board acknowledged that it is theoretically possible for the automated pricing mechanism to produce outcomes analogous to algorithmic hub-and-spoke risks, it stated that certain cumulative conditions must be satisfied for such a model to be assessable under Article 4. Specifically: a seller must provide the platform with its forward-looking pricing strategy in anticipation that it will be conveyed to other sellers; the platform must transmit that information; the seller receiving the information must be in a position to know the circumstances under which the originating seller provided it; and that information must be used in determining the recipient seller’s pricing strategy. The Board concluded that the requisite conditions had not been established in the specific case and that the mere use of an algorithm cannot in itself be treated as a concurrence of wills.

The Board recalled that algorithmic pricing does not constitute a by-object infringement in every case and that such tools may, under certain conditions, produce procompetitive efficiency gains. It further stated, however, that the potential of algorithms to facilitate tacit coordination by automating decision-making processes must be separately evaluated in the specific case.

As regards the specific mechanism at issue, the Board noted that sellers are able to set their prices by reference to benchmarks such as the “Competitive Featured Offer,” “Competitive Lowest Price,” and “External Competitor’s Price,” and that the Competitive Featured Offer price is dynamic and multi-parametric for each product. The Board found no direct or indirect contact among sellers with respect to competitive parameters, and concluded that the reference price is determined in accordance with criteria applied uniformly across all sellers within the platform.

Against this backdrop, the Board considered in aggregate the following factors: the automated pricing mechanism is not mandatory; no concurrence of wills constituting an agreement and/or concerted practice within the meaning of Article 4 among sellers with respect to use of the mechanism had been established; the rule set created under the automated pricing mechanism is, for the most part, designed to be customizable by each individual seller; each seller is able to set different validity periods for the rules it defines; and the automated pricing mechanism employs a rule-based rather than a learning-based algorithm. On this basis, the Board concluded that the existence of a competition-restricting agreement or concerted practice within the scope of Article 4 of Law No. 4054 had not been established in the specific case.

In addition to the foregoing assessments, Amazon made a notification on October 4, 2024, regarding the future of the automated pricing mechanism. Amazon stated that the automated pricing mechanism in its entirety, not limited to the rule set that is the subject of the investigation, would be disabled in Türkiye as of February 10, 2025. In its defense, Amazon argued that the alleged competitive concerns had been eliminated by shutting down the mechanism and that the investigation should therefore be closed. The Board, however, emphasized that, absent the operation of the commitment mechanism, a prospective operational change does not by itself bring about the automatic termination of the investigation with respect to allegedly anticompetitive conduct occurring in the past. The Board also noted in its decision that the said petition contained no explicit and binding commitment to the effect that the automated pricing mechanism would not be reactivated in the Turkish market in the future. Accordingly, the Board based its assessment not on the declaration that the mechanism would be shut down, but on the question whether the existence of a competition-restricting agreement or concerted practice within the scope of Article 4 of Law No. 4054 had been established in the specific case.

Conclusion

The Board’s assessment focused on the question of whether the automated pricing mechanism gave rise to an agreement or concerted practice among sellers within the meaning of Article 4 of Law No. 4054. In the final decision, it was concluded that sufficient evidence had not been established to demonstrate that forward-looking pricing intentions had been transmitted among sellers, or that coordination had been achieved, through the mechanism offered by Amazon.

The decision makes clear that algorithmic pricing tools provided by digital platforms do not, standing alone, constitute a competition infringement; an infringement finding requires concrete evidence establishing a concurrence of wills. The existence of theoretical risks analogous to the hub-and-spoke model is insufficient to prove that such an infringement occurred in the specific case.

The fact that the commitment route was taken with respect to the other platforms in the same investigation process, while an infringement finding was not reached in Amazon’s case, demonstrates that the Board conducts a separate, evidence-based assessment for each undertaking.

This decision constitutes an important point of reference in the assessment of algorithmic tools under Article 4 in digital markets, particularly with respect to the standard of proof and the analysis of coordination.

[1] Attorney Gülce Korkmaz is the external competition law consultant of Baş | Kaymaz Law Firm. After completing her master’s degree at Bilkent University, she is currently pursuing her doctoral studies in the field of competition law at the Faculty of Law of Lüneburg Leuphana University (Germany) as a PhD researcher with the scholarship of the Joachim Herz Foundation.
[2] For the reasoned TCB decision dated April 18, 2025, and numbered 25-15/348-164, please see here (only available in Turkish).
[3] TCB decisions dated October 3, 2024, and numbered 24-40/950-409 and 24-40/951-410.
[4] TCB decisions dated May 12, 2011, and numbered 11-30/591-187 (Gittigidiyor); January 3, 2013, and numbered 13-01/7-7 (D&R); November 10, 2015, and numbered 15-40/662-231 (Hepsiburada); June 9, 2016, and numbered 16-20/347-156 (Yemeksepeti); January 5, 2017, and numbered 17-01/12-4 (Booking); and October 1, 2018, and numbered 18-36/584-285 (Sahibinden). Turkish Competition Authority (2022), Final Report of the Sector Inquiry into E-Marketplace Platforms, https://www.rekabet.gov.tr/Dosya/geneldosya/e-pazaryeri-si-raporu-pdf.
[5] CJEU, Eturas UAB and Others v. Lietuvos Respublikos konkurencijos taryba, Case No. C-74/14, January 21, 2016, ECLI:EU:C:2016:42; [2016] OJ C 98/3.
[6] See https://en.kfst.dk/nyheder/kfst/english/decisions/20200630-danish-competition-council-ageras-has-infringed-competition-law/; https://kfst.dk/media/ws5nbdtx/20200630-ageras-final-a.pdf.
[7] CMA, Case 50223, https://assets.publishing.service.gov.uk/media/57ee7c2740f0b606dc000018/case-50223-final-non-confidential-infringement-decision.pdf.
[8] Superior Court of the District of Columbia, Case No. 2023 CAB 6762, https://business.cch.com/ald/DistrictofColumbiavRealPageInc792024.pdf.
[9] Spencer Meyer v. Travis Kalanick, 15 Civ. 9796; 2016 U.S. Dist. LEXIS 43944.

As Baş | Kaymaz Law Firm, we provide comprehensive competition law, advice and representation services to both national and multinational companies with our specialized and experienced lawyers.

If you have any questions on this topic or any matter related to Turkish competition law, you may contact us via [email protected].