To quote this paper: L. GESTIERO, Two-sided markets and evidence in competition law : The Autority’s dismissal of the case due to the lake of conclusive evidence of abuse of a dominant position and economic dependance in research syndication, Competition Forum, 2026, n° 0086 https://competition-forum.com. 

By Decision No. 25-D-08 of November 27, 2025, the Competition Authority (hereinafter “the Authority”) dismissed the complaint filed by Qwant on the merits and, consequently, the related request for interim measures, due to a lack of sufficiently compelling evidence. This decision is consistent with established case law, according to which competition law is, above all, a matter of evidence and rigorous economic analysis.

Qwant, a French search engine that prioritizes privacy protection, has been using Microsoft’s search syndication services since 2016. These services involve providing a third party with organic search results and search-related advertisements (paid results). In its complaint, Qwant alleged that Microsoft engages in exclusivity and tying practices, imposes restrictions on the development of artificial intelligence services, and engages in discriminatory practices regarding access to advertising, particularly through the “Select” status on the Microsoft Advertising platform.

These practices are alleged to have constituted an abuse of a dominant position within the meaning of Article 102 of the Treaty on the Functioning of the European Union (TFEU) or, failing that, an abuse of economic dependence within the meaning of the second paragraph of Article L. 420-2 of the Commercial Code. The decision concerns a sector characterized by massive fixed costs (crawling, indexing, ranking algorithms), cross-network effects between search and advertising, and significant economic concentration.

From a methodological standpoint, the Authority notes that, even before discussing the legality of a practice, any referral must establish a credible relevant market as well as a position of economic power[1]. The evidentiary requirement, far from constituting excessive formalism, is a sine qua non condition ensuring that the analysis is not based on a mere intuition of contractual unfairness.

The central question raised by the decision can thus be formulated as follows: under what evidentiary conditions can a company effectively invoke competition law to challenge contractual terms and practices in a two-sided digital ecosystem, where economic power is assessed in the shadow of “ultra-dominant” players? The Authority answers this by successively ruling out the classification of abuse of a dominant position (I), followed by that of abuse of economic dependence (II).

I. Rejection of the Alleged Abuse of Dominance: The Burden of Proof in the Context of Multifaceted Markets

The Authority’s analysis follows the classic three-step approach in antitrust law: definition of the relevant market, assessment of market power, and, if applicable, determination of the abuse. In this case, the referral fails at the very first two stages, due to an insufficient definition of the relevant market (A) and the impossibility of establishing Microsoft’s dominance in the face of Google’s structural pressure (B).

A. An inadequate definition of the relevant market: a failure right from the “foundation” of the competitive analysis

The definition of the relevant market is a prerequisite for any analysis of a dominant position[2]. As the Court of Justice noted in the United Brands judgment, a dominant position is defined as ‘A position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market.’[3] This definition presupposes a precisely defined market. Qwant argued that there was a specific market for “search engine syndication,” in which only Bing’s offering could meet the demand from alternative search engines, due to linguistic barriers, performance issues, and technical limitations of competing offerings[4]. However, the Authority notes, on the one hand, that this definition has never been established by decision-making practice or case law[5], and on the other hand, that syndication is a heterogeneous mechanism serving players with very different needs (search engines, website publishers, AI services for grounding).

This analysis is consistent with the insights of the two-sided market theory developed by Rochet and Tirole[6] : syndication cannot be viewed as a single market, as its economic equilibrium depends on the advertising side. The Authority acknowledges, however, that it cannot be ruled out that syndication may constitute a market distinct from general[7] search, since the Commission had already noted that syndicated search engines were full-fledged providers in the general search market. But even under this assumption, the market would necessarily be multi-sided due to the cross-network effects between the syndication side and the advertising side.

Regarding the geographic scope, the Authority has adopted a national definition, consistent with the Commission’s previous decisions in the Google Search (Shopping) and Google Android cases, on the grounds that the online search and advertising markets are segmented by language barriers[8].

Ultimately, in the absence of quantified and documented evidence sufficient to “pin down” a relevant market, the Authority dismisses the referral for lack of supporting evidence, noting that the investigative services are not intended to make up for the referring party’s failure to provide such evidence[9]. This decision is consistent with established case law: the absence of sufficiently compelling evidence at the stage of market definition is sufficient grounds for rejecting the referral.

B. The Absence of Microsoft’s Dominance: Competitive Pressure from Google Undermines the Hypothesis of Market Power

For analytical purposes, the Authority nevertheless addresses Microsoft’s position in the relevant market. It notes, in accordance with the Hoffmann-La Roche ruling, that a dominant position may result from several factors considered in isolation, which are not necessarily decisive[10]. Qwant argued that Microsoft holds a dominant position in search syndication, on the grounds that only Bing offers a solution that allows a syndicated search engine to operate viably. The Authority acknowledges that Microsoft is the leading player in content syndication to alternative search engines, for which it is the provider (Yahoo, DuckDuckGo, Ecosia, Qwant), which may constitute a qualitative indicator of market power. However, an indicator is not a conclusion: dominance must be demonstrated by taking into account the overall competitive structure, countervailing forces, and substitutability[11].

The cornerstone of this reasoning lies in the multifaceted approach to the market. With regard to the advertising side, the Authority reiterates Google’s position of overwhelming dominance, as established by the Commission’s Google Search (Shopping) and Google Search (AdSense) decisions: in 2016, Google held [80–90%] of the market share in the EEA and [90–100%] in France for search-related advertising, while Microsoft’s share there was negligible. However, this dominance in the advertising market is not without impact on the syndication market: due to cross-network effects, Google is able to offer syndicated search engines a greater quantity and higher quality of advertisements than Microsoft.

The Authority explicitly describes these network effects: the more users a syndicated search engine attracts, the more the provider can improve the relevance of its ad targeting; conversely, a larger pool of advertisers increases the appeal of the syndication solution. In this context, Google’s dominance in the advertising market, reinforced by its dominant position in general search, fuels the syndication market by giving it a systemic advantage (ad inventory, performance, and appeal to advertisers).

Consequently, in the absence of solid factual evidence demonstrating that Microsoft would nevertheless be capable of acting independently of this pressure, the assumption that Microsoft holds a dominant position appears incompatible with the principles established by case law[12]. The legal consequence is immediate: in the absence of dominance, there is no need to examine the practices under Article 102 TFEU and Article L. 420-2, first paragraph of the Commercial Code[13].

This solution nevertheless raises an interesting legal question: in a digital ecosystem where Google is the overwhelmingly dominant player, could the pressure it exerts on Microsoft, paradoxically, “whitewash” the latter’s practices toward partners in a position of structural weakness? The decision does not answer this question, but it illustrates the limitations of antitrust law in two-sided markets with significant asymmetry, where the Digital Markets Act (DMA)[14] could provide complementary tools.

II. Rejecting the Abuse of Economic Dependence: Contractual Balance and Insufficient Evidence

Article L. 420-2, second paragraph of the Commercial Code provides a complementary tool to the law on dominant market positions, allowing for the punishment of the abusive exploitation of a state of economic dependence without the need to demonstrate market power in the strict sense. However, the classification of abuse of economic dependence is based on the fulfillment of three cumulative conditions: the existence of a situation of dependence, the abusive exploitation of that situation, and an impairment of the functioning or structure of competition. In this case, dependence is not established (A), and the alleged practices do not reveal any abusive exploitation (B).

A. Unestablished economic dependence: alternatives and Qwant’s path to autonomy

Economic dependence is assessed on a case-by-case basis according to several objective criteria established by case law[15]: the proportion of revenue generated with the partner, the size of its market share, its reputation, and, above all, the absence of technically and economically equivalent alternatives. The Paris Court of Appeal has clarified that the impossibility of an equivalent solution refers to both the legal and practical ability to contract otherwise[16].

About the criterion of external alternatives, the Authority acknowledges that Microsoft’s offer has specific features not found in certain competing offers (the absence of advertising in Google’s mobile search results, and the inferior quality of results from Brave and Mojeek), and that the substantial portion of Qwant’s revenue derived from Microsoft’s syndication is a significant factor. However, these findings are not sufficient to establish a lack of alternatives: they demonstrate a difference in functionality, not a total absence of solutions.

About the criterion of internal alternatives, the Authority highlights Qwant’s unique situation: for the past fourteen years, the company has been developing its own search technology and already processes approximately half of its search queries in France using its own capabilities; furthermore, it is the only Microsoft partner contractually permitted to develop its own technology. The creation of the European Search Perspective entity with Ecosia and the launch of the “Staan” syndication offering in June 2025 illustrate a trend toward autonomy that is transforming Qwant into a fully-fledged syndication provider.

However, Qwant is not unable to act otherwise, but rather faces a costly industrial and economic choice, which is not the same thing from a legal standpoint. The Authority, in line with its previous decision-making practice[17], concludes that the allegations of dependence are not supported by sufficient evidence.

Beyond this specific case, this solution raises questions for legal scholars regarding the relevance of the dependency threshold in digital ecosystems: lock-in can be gradual and structural, resulting not from a legal constraint but from a diffuse economic constraint linked to network effects and switching costs. 

B. Absence of exploitative conduct: economic justification of the practices and insufficient evidence

Even assuming that dependency had been established, it would still be necessary to demonstrate exploitative conduct. The Authority examines the three alleged practices in turn and dismisses them all on the dual grounds of objective economic justification and insufficient evidence.

1. Advertising exclusivity and tied sales: the logic of contractual balance

Regarding exclusivity and tied sales, the Authority adopts an interpretation grounded in contract economics. Exclusivity serves as the consideration for preferential terms granted to Qwant in a model where the provision of organic search results is nearly at marginal cost, offset by advertising monetization.

The Authority emphasizes that, in the absence of exclusivity, partners could offer the same advertising inventory to both Google and Microsoft simultaneously. The Google platform, which is more attractive due to its dominance in search-related advertising, would then capture the bulk of advertising spending, further weakening Microsoft’s already marginal position. This analysis is consistent with the economic rationale underlying exclusivity agreements in bilateral markets: as Evans and Schmalensee note, exclusivity can, in certain contexts, serve as a mechanism enabling a platform to secure the simultaneous presence of both sides of the market and achieve the critical mass necessary for its economic viability, without this constraint necessarily constituting an anticompetitive restriction.[18] Finally, the Authority notes that Microsoft offers the option to enter into separate contracts for organic results alone or advertisements alone, under different pricing terms, and that Qwant did not choose this option.

2. The restriction regarding artificial intelligence: a condition for legitimate use

Regarding artificial intelligence, the Authority makes a decisive distinction: it differentiates between the restriction alleged by Qwant (the inability to develop AI models) and the restriction actually found (the prohibition on using syndication results to train or perform inference in chatbots). The former does not exist contractually; the latter is deemed legitimate.

This distinction is legally relevant: since no advertising links are displayed in a chatbot, Microsoft receives no compensation for the use of its data, which justifies a separate agreement for this purpose. The Authority notes that Microsoft does indeed offer separate contracts for AI clients, which Qwant has not signed and whose legality it does not dispute. The restriction thus appears to be a condition of use consistent with the partnership’s compensation structure, rather than an anticompetitive restriction aimed at protecting an advantage in the deployment of generative AI.

3. Discrimination in access to advertising: lack of evidence of anti-competitive conduct

Regarding advertising discrimination, the Authority notes that Microsoft ranks its partners’ traffic based on four performance criteria (conversion rate, traffic volatility, cost per acquisition, dwell time), which determine eligibility for the “Select” category and more lucrative ads. Qwant alleged that these criteria were non-objective, non-transparent, and applied in a discriminatory manner.

The Authority rejects this allegation on several grounds. First, the criteria correspond to metrics commonly used to assess the value of traffic for advertisers. Second, the number and highly detailed nature of the exchanges between Microsoft and Qwant regarding the functioning of the classification system refute the allegation of a lack of transparency. Third, privacy-focused search engines, by definition, provide less targeting data, which reduces the relative value of their traffic to advertisers, without this resulting from discrimination attributable to Microsoft.

This approach is consistent with the principles established in case law on discrimination under competition law[19]: a decline in advertising revenue can result from a variety of factors (traffic mix, query quality, advertiser behavior, economic conditions) and does not, on its own, prove anti-competitive discrimination. Qwant has not provided sufficient detailed evidence to demonstrate that the classification criteria, in and of themselves or in their implementation, constitute an abuse of a dominant position within the meaning of Article L. 420-2, second paragraph, of the Commercial Code.

Decision No. 25-D-08 is as much a procedural decision as it is a substantive one. It reiterates that, in digital markets, economic sophistication, network effects, two-sided markets, and the integration of advertising and search do not lessen the burden of proof: they reinforce it. By refusing to find Microsoft dominant, the Authority highlights a fundamental phenomenon: the analysis of market power cannot be separated from the revenue-generating side of the market in this case, advertising, which is dominated by Google whose competitive pressure reshapes the balance of the entire content distribution ecosystem.

By setting aside issues relating to economic dependence and alleged abuses, the Authority focuses its analysis on the contractual balance and the objective justification for the disputed clauses, emphasizing the lack of evidence of any harm to the structure or functioning of competition. This approach is consistent with the case law of the Court of Justice, which, in the TeliaSonera case, requires proof of an anticompetitive effect even if only potential to establish the abusive nature of a foreclosure practice.[20]

Beyond the specific case at hand, this decision raises several doctrinal questions of general significance. First, there is the question of the limits of the law on dominant positions in highly asymmetric two-sided markets, where the presence of an ultra-dominant player (Google) can paradoxically reduce the ability of other players (Qwant) to challenge the practices of an intermediary player (Microsoft). Second, the question of the relationship between competition law and the DMA[21], which designates Microsoft as the gatekeeper for Bing, Edge, and Windows, and could provide intervention tools better suited to the constraints of digital markets than the traditional instruments of antitrust and abuse of dominance law.

Lucie GESTIERO

[1] CJUE, February 14, 1978, United Brands and United Brands Continental BV v. Commission, Case 27/76, paragraphs 65-68.

[2] Communication from the European Commission on the definition of the relevant market for the purposes of EU competition law, OJEU C 2024/1645, February 22, 2024.

[3] CJUE, February 14, 1978, United Brands and United Brands Continental BV v. Commission, Case 27/76, paragraph 65.

[4] ADLC, dec. N°25-D-08,  November 27, 2025, paragraph 63.

[5] ADLC,  dec. N° 25-D-08, November 27, 2025, paragraph 64.

[6] Tirole & J.-C. Rochet, « Platform Competition in Two-Sided Markets », Journal of the European Economic Association, 2003, vol. 1(4), p. 990-1029.

[7] Commission Decision of June 27, 2017, Google Search (Shopping), COMP/AT.39740, upheld by the General Court’s judgment of November 10, 2021, Google and Alphabet v. Commission (Google Shopping), T-612/17, EU:T:2021:763, and subsequently by the judgment of the Court of Justice of September 10, 2024, Google and Alphabet v. Commission (Google Shopping), C-48/22 P.

[8] Competition Authority, Decision No. 19-D-26 of December 19, 2019, regarding the search-based online advertising sector, para. 300 et seq.; Decision No. 20-MC-01 of April 9, 2020 (Google/neighboring rights), paras. 146 et seq.

[9] Administrative Decision No. 25-D-01 of February 19, 2025, regarding the civilian explosives sector, paragraph 54 referring to the decisions of the Court of Cassation, Commercial Chamber, of January 19, 2016, appeals Nos. 14-21670 and 14-21671, as well as Decisions No. 16-D-29 of December 19, 2016, concerning practices in the after-sales sector for reprographic equipment (paragraphs 19 and 20), No. 18-D-20 of October 5, 2018, concerning practices in the sector of publishing and marketing IT management solutions for the agricultural sector (paragraphs 28 and 29), No. 19-D-03 of January 16, 2019, concerning practices in the sector of cross-Channel transport of day-old chicks (paragraphs 31–39), No. 21-D-18 of July 15, 2021, concerning practices in the sector of unsold newspaper and magazine management (paragraphs 26–34), and No. 22-D-07 of February 23, 2022, concerning practices in the sector of the production and distribution of audiovisual content via video-on-demand (paragraphs 20–26).

[10] ECJ, February 13, 1979, Hoffmann-La Roche & Co. AG v. Commission, Case 85/76, para. 48.

[11] General Court of the European Union, November 10, 2021, Google LLC v. Alphabet, T-612/17, para. 155.

[12] ECJ, February 13, 1979, Hoffmann-La Roche & Co. AG v. Commission, Case 85/76, para. 38.

[13] ECJ, November 9, 1983, NV Nederlandsche Banden Industrie Michelin v. Commission of the European Communities, Case 322/81 paragraphs 29-31.

[14] Regulation (EU) 2022/1925 of the European Parliament and of the Council of September 14, 2022, on contestable and fair markets in the digital sector (DMA), October 12, 2022.

[15] Competition Authority, Decision No. 01-D-49 of August 31, 2001; Consumer Affairs, Decision No. 10-D-08 of March 3, 2010, regarding the local general food retail sector.

[16] Paris Court of Appeal, Paris, October 6, 2022, Case No. 20-08582, paragraphs 533.

[17] Competition Authority Decision No. 18-D-25 of December 6, 2018, regarding the takeout and pizza delivery sector, paragraphs 74 and 75.

[18] Evans, David S. and Schmalensee, Richard (2010) “Failure to Launch: Critical Mass in Platform Businesses,” Review of Network Economics: Vol. 9 : Iss. 4, Article 1, p.23.

[19] General Court of the European Union, December 17, 2003, British Airways v. Commission, T-219/99, para. 65.

[20] CJEU, TeliaSonera Sverige Case, C-52/09, paragraph 64.

[21] Regulation (EU) 2022/1925 of the European Parliament and of the Council of September 14, 2022, on contestable and fair markets in the digital sector (DMA), October 12, 2022.