Does the competition authority have to prove that an equally efficient competitor is being excluded?

In cases involving the abuse of a dominant position, the same evidentiary question arises time and again: must the regulatory authority demonstrate that the conduct excludes an equally efficient competitor, and must it develop a counterfactual scenario? In its July 2, 2026, ruling in the Google Android case (C-738/22 P) The Court of Justice answers both questions in the negative. Neither the “equally efficient competitor” test nor a counterfactual analysis is required in all cases to establish a violation of Article 102 TFEU. Google and Alphabet had their appeals dismissed, and the fine of approximately 4.1 billion euros was upheld.

The facts

The case dates back to A decision of the European Commission dated July 18, 2018, in which it found that Google had abused its dominant position in the markets for general search services through a series of contractual terms tied to the Android operating system.

There were three categories of restrictions. First, the distribution agreements (MADAs), under which mobile device manufacturers were required to pre-install the Google Search app and the Chrome browser in order to obtain a license for the Play Store. Second, the anti-fragmentation agreements (AFAs), under which those manufacturers committed not to sell devices running versions of Android not approved by Google (so-called “incompatible Android forks”). Third, the revenue-sharing agreements (RSAs), under which manufacturers and network operators received a share of advertising revenue on the condition that they did not pre-install competing search services.

The Commission considered these restrictions to constitute a single continuous infringement due to their common purpose and interconnection, and imposed a fine of 4,342,865,000 euros. In 2022, the General Court upheld that classification but annulled the decision to the extent that it characterized the portfolio-based RSAs as an abuse, and reduced the fine to 4,125,000,000 euros. Google and Alphabet filed an appeal against that judgment.

The decision

The Court of Appeals dismisses the appeal in its entirety and upholds the fine as revised by the District Court.

The contextual value of non-prohibited conduct

Google criticized the General Court for failing, in its assessment of the pre-pre-installation conditions in the MADAs, it had taken into account the effects of the RSAs, even though the portfolio-based RSAs had not been classified as abusive and the device-based RSAs were entirely outside the scope of the decision.

The Court does not follow that line of reasoning. Whether a factual element can be taken into account as a relevant contextual factor in the assessment under Article 102 TFEU does not depend on whether that element in itself constitutes conduct that qualifies as an abuse. The key issue is whether it is possible to assess, in concrete terms, whether the effects of a conduct are amplified or mitigated by the context in which it occurs. The General Court had found, as a matter of fact, that the MADAs and the RSAs were complementary and that their effects could not be viewed in isolation from one another. An artificial distinction between the two would make it impossible to demonstrate the unlawfulness of conduct on the basis of precise and concrete elements, particularly in markets with network effects.

No mandatory testing of the equally efficient competitor

The core of the second ground of appeal concerned the “as-efficient competitor” test. Google argued that the Commission, just as it had done with regard to the RSAs, should have demonstrated that the pre-installation conditions could exclude an equally efficient competitor.

The Court rejects that premise. To establish abuse, it must be proven that the conduct is based on means other than competition on the merits, and that it may have exclusionary effects. Both conditions are cumulative, but neither systematically requires the application of the “equally efficient competitor” test. For certain types of conduct, such as tying, the classification generally falls within the category of practices that deviate from competition on merit without the need for such an analysis. Furthermore, there are market structures in which that test is neither possible nor useful: in a digital ecosystem with high barriers to entry and network effects, a particular conduct may, in practice, make it impossible for an equally efficient competitor to enter the market or sustain its position.

No mandatory counterfactual analysis

Google also criticized the General Court for having confirmed the effects of the anti-fragmentation agreements without developing a counterfactual scenario. Here too, the Court of Justice held that the Commission must demonstrate the causal link between the conduct and the anticompetitive effects, but that it may rely on a body of evidence to do so and is not required to systematically use a single method, in particular a counterfactual analysis. Since the existence of the restrictions and their anticompetitive effects had been sufficiently demonstrated, such an analysis was not necessary in this case.

The reclassification of the conduct and the partial annulment

The Court confirms two additional points. The General Court had not “rewritten” the finding of abuse regarding the anti-fragmentation agreements: the differing wording referred to the same conduct as that described in the decision, namely preventing incompatible Android forks from finding markets. Furthermore, the partial annulment of the decision regarding the portfolio-based RSAs did not preclude maintaining the classification as a single ongoing infringement, since the remaining conduct continued to form part of the same overall strategy.

Legal analysis and interpretation

Confirmation of the effects-based approach, but with flexible standards of proof

The general thrust of this ruling is consistent with recent case law, in particular the ruling of September 10, 2024, in the Google Shopping case (C-48/22 P). The Court reiterates that an exclusionary practice constitutes an abuse only if both a deviation from competition on merit and the possibility of exclusionary effects are demonstrated. This is an effects-based approach.

The innovation lies in the burden of proof. The Court puts into perspective two analytical tools that were often presented as necessary in practice: the equally efficient competitor test and the counterfactual analysis. Both are reduced to possible, but not mandatory, methods. For digital ecosystems, the Court goes even further: there, the equally efficient competitor test may be downright unsuitable, because the market structure makes it impossible for a competitor to sustain itself in any case. This reasoning is coherent, but at the same time it lowers the burden of proof for the regulator. Anyone who characterizes the conduct as deviating from competition on merit and demonstrates the possibility of exclusion no longer needs to prove, using a quantitative model, that a hypothetical equally efficient competitor would be driven out of the market.

There is a tension here. The recognition that some markets do not lend themselves to the “equally efficient competitor” test is defensible. But there is a risk that “the special characteristics of the digital economy” will become a category in which the dominant firm has structurally less basis for having its conduct reviewed in advance. The Court partially addresses this by continuing to require that the possibility of exclusion be demonstrated on the basis of precise and concrete analysis and evidence, rather than in the abstract.

The contextual relevance of non-prohibited conduct broadens the framework for assessment

Perhaps the most far-reaching point of principle is that conduct that has not itself been classified as abuse may nevertheless be taken into account as a contextual factor when assessing the conduct that is actually being targeted. The Court separates the relevance of a factual element from the question of whether that element is unlawful in and of itself.

This is defensible from a doctrinal perspective: the assessment of whether conduct constitutes an abuse cannot be made in the abstract, and the combined effect of multiple agreements is a real market reality. Nevertheless, this approach significantly broadens the framework for assessment. Lawful commercial agreements can contribute to the unlawfulness of other agreements without having to meet the requirements of Article 102 TFEU themselves. This is a point of concern for the legal certainty of dominant firms: the lawfulness of an agreement that is permissible in and of itself offers no guarantee that it will remain outside the scope of competition law analysis.

Specifically, what does this mean?

For companies that may hold a dominant position. The lesson is that the validity of an individual agreement is not sufficient to avoid scrutiny. The combined effect of various contractual mechanisms is assessed in light of their interrelationships, even when some of them are perfectly permissible on their own. Anyone who builds their distribution and licensing practices on a web of mutually reinforcing conditions must assess that architecture as a whole, not clause by clause. This is all the more true in digital markets with network effects, where the ability to rely in advance on the “equally efficient competitor” test is no longer guaranteed.

For competitors and customers who have been harmed. This ruling does not increase the burden of proof on the regulator; in fact, it makes it slightly more flexible in certain respects. For companies that suffer harm as a result of exclusionary conduct, this means that a complaint filed with the competent authority does not necessarily have to be based on an economic model of the “equally efficient competitor” type. Precise and concrete evidence regarding the functioning of the market and the possibility of exclusion is sufficient. Furthermore, anyone considering a claim for damages can rely on a decision that has become final following this ruling, with the customary evidentiary weight accorded to it in follow-on proceedings.

Frequently asked questions (FAQ)

Must a competition authority always apply the “equally efficient competitor” test in cases of abuse of a dominant position?
No. The Court of Justice confirms that this test is one of several possible methods of analysis, but it is not required in all cases. For certain types of conduct, such as tying, and certainly in digital markets with high barriers to entry, the test may even be inappropriate.

Does the Commission need to develop a counterfactual scenario to demonstrate abuse?
No. While the Commission must prove the causal link between the conduct and the anti-competitive effects, it may rely on a body of evidence to do so. A counterfactual analysis is not systematically required.

Can a valid agreement still play a role in determining whether abuse has occurred?
Yes. Conduct that is not itself classified as abuse may be taken into account as a relevant contextual factor in the assessment of the conduct in question, particularly when the consequences of both are inextricably linked.

Conclusion

With this ruling, the Court of Justice reaffirms the effects-based approach to abuse of a dominant position, but at the same time clarifies that the regulatory authority is not bound to a single, fixed method of proof. Neither the “equally efficient competitor” test nor a counterfactual analysis is required in all cases, and even conduct that is lawful in and of itself may be taken into account in the broader context. For companies with market power—and certainly for players in digital ecosystems—this means that the competition law risk analysis is shifting from individual clauses to the overall set of mutually reinforcing practices.