by Ece Ban and Anush Ganesh
On 3 September 2026, the European Commission adopted its Guidelines on the application of Article 102 TFEU to abusive exclusionary conduct by dominant undertakings (‘the Guidelines’). Replacing the 2009 Guidance on the Commission’s Enforcement Priorities, the Guidelines aim to contribute to the modernisation of competition policy, enhance legal certainty, and bring greater consistency to Article 102 TFEU enforcement (see the press release). In this blog, we look at what has changed from the Draft Guidelines on exclusionary abuses of dominance (‘the 2024 Draft’), the key elements retained, and the gaps that remain.
The promise and the scope of the Guidelines
The Guidelines seek to bring greater clarity and predictability to Article 102 TFEU enforcement (para 7). This ambition to clarify the current state of case law sets a different tone from the 2009 Guidance Paper, which aimed to set the Commission’s enforcement priorities. In this context, it provides a comprehensive summary of current case law, making it a valuable instrument for stakeholders, academics, and practitioners. Going even further than the 2024 Draft, the Guidelines state that their scope extends beyond summarising the current state of case law to explaining the Commission’s position on issues the courts have not yet addressed or that remain open to interpretation (para 8). This forward-looking guidance is welcome and further strengthens the Guidelines’ legal certainty objective. Yet, at times, the distinction between these two tasks becomes unclear in specific paragraphs; clearer signposting would have clarified where the Commission relies on the second task and steps beyond case law (thereby more exposed to being tested by the court). The Guidelines also recognise the dynamic nature of case law. It states it will further develop by taking into account developments in the case law and adds that the Commission may depart from the Guidelines to keep enforcement practice aligned with later developments in the case law (fn 17).
Dominance analysis
The treatment of dominance largely follows the 2024 Draft, but the final version adds a few key points. First, the Guidelines bring back the soft safe harbour for dominance (para 24) from 2009 Guidance (see para 14 of 2009 Guidance), which was dropped in the 2024 Draft. Drawing on its enforcement experience, the Commission states that dominance is generally unlikely below market shares of 40% (para 24). Yet, it can still be established below that level (e.g. where customers depend heavily on the undertaking or competitors face serious capacity constraints) (para 24).
Second, the discussion of barriers to entry has been expanded compared to the 2024 Draft, with more emphasis on entry barriers in the digital economy. While the 2024 Draft stated that data-driven advantages can act as an entry barrier, the Guidelines specifically exemplify factors that such advantages may arise from: access to unique or non-replicable data, the accumulation of data, data-related economies of scale or scope, network effects, and the capabilities to process and utilise data (para 31). The Guidelines also state that these factors are especially relevant for artificial intelligence, where access to datasets and computing power is crucial (para 31). The Guidelines also mention network effects in digital ecosystems (a concept not included in the 2024 Draft) and state that in such markets, user lock-in, high switching costs, and data-driven advantages that may relate to multiple products may make rivals’ entry and expansion difficult when undertaking holds a dominant position in one or more markets that are part of an ecosystem (para 34).
Third, the Guidelines devote greater attention to collective dominance and place additional emphasis on the role of algorithmic systems in facilitating tacit coordination. The Commission recognises in the Guidelines that the growing use of algorithms may make findings of collective dominance more common by facilitating coordination even in complex environments (fn 105). More specifically, it points out that algorithms may make it easier for firms to reach coordination terms and make otherwise less transparent markets easier to monitor (paras 50 – 51).
The expansion of the section on data-driven entry barriers, the appearance of the concept of digital ecosystems, as well as the role of algorithms in facilitating tacit coordination, show that the Commission’s final version brings the dominance analysis more in tune with the realities of the digital economy – a welcome clear shift from the 2024 Draft.
Finding abuse: the general framework remains unchanged
The two-limb test for abuse, and how it operates in practice, attracted considerable attention in the debate over the 2024 Draft. The Guidelines retain this case-law-derived conceptual structure, stating that conduct ‘distorts effective competition’ where it (i) departs from competition on the merits, (ii) is capable of having exclusionary effects (para 59).
Going beyond the 2024 draft, the Guidelines further clarify that the test is not sequential and that there is not necessarily strict separation between these prongs; at times, certain parts of the analysis and the same evidence will be relevant to both (para 61). The Guidelines then identify three situations in which analysis of the two-limb test is unnecessary: (i) conduct governed by a specific analytical framework; (ii) conduct capable of excluding a hypothetical equally efficient competitor; and (iii) conduct by its very nature is harmful to competition (paras 62-65). Conduct that does not fall under these three categories is subject to the general conceptual two-limb test (para 59). Before looking at these specific categories, we turn to the most controversial element of the two-limb test: ‘competition on the merits’, and how the Guidelines conceptualised the concept.
Article 102 TFEU’s ‘42’: competition on the merits
The Guidelines define competition on the merits as based on ‘performance of the economic operators’ (para 67), retaining the definition used in the 2024 Draft. It further states that this in principle covers ‘a competitive situation in which consumers benefit from lower prices, better quality and a wider choice of new or improved products, both in the short and long term’ (para 67). It also states that conduct having the effect of ‘broadening consumer choice by putting new goods on the market or by increasing quantity or quality of the goods on offer generally falls within the scope of competition on the merits’ (para 67). It also clarifies that competition on the merits concerns not only conduct but also the economic and legal context (para 70). These formulations closely track existing case law and are drawn from it. Yet read together, the overarching definition of competition on the merits seems to shift between ideas, conflating performance and outcomes (for discussion regarding the problematic aspect of the latter, see this consultation response, page 8).
Rather than a negative conceptual definition, the Guidelines provide non-exhaustive relevant factors to establish that conduct departs from competition on the merits (paras 70-78). While the 2024 Draft listed all factors as ‘relevant’, the Guidelines now seem to distinguish them by their weight. The first category of factors is listed as providing a ‘strong indication’ of a departure from competition on the merits (para 72). These include providing misleading information to public authorities, misusing regulatory or legal procedures to impede entry, and violating rules in other areas of law in a way that negatively affects a parameter of competition (para 72). The second category consists of factors constituting an ‘important factor’ in determining whether conduct departs from competition on the merits where undertakings’ conduct ‘relies on the use of resources or means inherent to the holding of the dominant position’ (para 73). The Guidelines then list other (non-exhaustive) ‘relevant’ factors, including whether the conduct prevents consumers from exercising choice based on merits or involves biased or discriminatory treatment that favours the dominant undertaking over its competitors (para 74). While the differentiated language creates a hierarchy of evidential weight, its practical implications are left unclear.
On ‘competition on the merits’, the Guidelines appear to offer little beyond the 2024 Draft and did not answer the calls for a clearer general principle made during the consultation. The status of ‘competition on the merits’ brings to mind the Hitchhiker’s Guide to the Galaxy. In Douglas Adams’s sci-fi novel, the supercomputer Deep Thought finds that cryptic answer to the ultimate question of life, the universe and everything is ‘42’. Yet the answer leaves the creators unsatisfied, as they still do not know the question to be asked to get to the answer. ‘Competition on the merits’ has long risked becoming Article 102 TFEU’s ‘42’: a departure from competition on the merits has been articulated as a general principle for identifying conduct as abuse, but the questions to be asked to reach the answer remain elusive. Indeed, the abstract nature of ‘competition on the merits’ has long been seen as a weakness (see, for example, AG Rantos Opinion, Servizio Elettrico Nazionale, paras 55- 57; OECD’s Background Note in 2005, which stated that ‘no one really knows what [it] means’). With ‘competition on the merits’ articulated as a limb of the general conceptual test, giving it operational meaning is more crucial than ever to future-proofing Article 102 TFEU and ensuring legal certainty. As an overarching principle, it will guide the assessment of novel conduct for which no specific analytical framework yet exists, and that does not fall under other categories mentioned (paras 62-65). While we have more signposts, the questions to be asked to operationalise the concept remain ambiguous beyond the few relevant, non-exhaustive factors mentioned.
Further clarifications on general principles on finding abuse: theory of harm, sliding scale and causation
The Guidelines also contain three additions, different from the 2024 Draft, that are worth looking into more closely. The first is the definition of the ‘theory of harm ‘, a concept that has been (widely and divergently) used in the context of Article 102 TFEU enforcement. Accordingly, the Guidelines state that the theory of harm refers to ‘the economic mechanism through which the conduct is capable of directly or indirectly harming consumers, in its specific economic and legal context’ (para 57).
Second, the Guidelines set out a sliding scale: ‘the more a given conduct is considered generally likely to distort effective competition, the less case-specific evidence is required to prove that this is the case’ (para 58). The paragraph does not cite any judgment, making it clear that this part comes within the scope of the Commission’s interpretative task. Yet the principle is anchored in the case law and embedded in the design of different specific cases applicable to different conduct. Indeed, the Commission illustrates this point by contrasting the legal test applicable to self-preferencing and exclusive dealing (fn 124).
Third, the Guidelines devote a section to causation. While causality between effects and conduct has been well discussed, it also mentions ‘conduct causality’ (causality between dominance and the conduct), which has received relatively little discussion in case law (though it is increasingly discussed in recent Commission decisions on Apple App Store Practices and Facebook Marketplace). The Guidelines, by citing older case law Hoffman La Roche and Continental Can, state that whether conduct in question is capable of producing exclusionary effects does not require that the conduct is enabled by dominant position (para 99, see also para 76). On the other hand, relying on resources from a dominant position is a relevant factor in finding that conduct departs from competition on the merits (para 73).
Predatory pricing: Cost proxies and two-sided platforms
The Guidelines define predatory pricing as a below-cost pricing practice aimed at marginalising competitors or otherwise reducing competition (para 108). This is a narrower formulation than the one used in the 2009 Guidance Paper, which framed the same conduct around the deliberate sacrifice of profitability (para 63 of the 2009 Guidance). One commentator has argued that the change is substantive as it removes the possibility of establishing predation where an undertaking prices above cost while deploying capital inefficiently or accepting a depressed rate of return.
The operative test keeps the AKZO structure intact (para 113). Pricing below average variable cost (AVC) is considered predatory on the reasoning that every sale generates a loss and no objective other than the elimination of competitors can sensibly be attributed to it (para 113a, C-62/86, AKZO para 71). Pricing between AVC and average total cost (ATC) becomes predatory only where it forms part of a plan to eliminate or reduce competition, with that plan demonstrable through direct evidence, indirect evidence, or a combination of the two provided it is sound and consistent (para 113b, T-340/03, France Télécom para 197). Pricing above average total cost is not predatory at all. Recoupment remains irrelevant to liability, though the Commission retains the ability to treat it as evidence bearing on the assessment (para 109, C-209/10, Post Danmark, para 36).
On costs, taking a leaf out of the Qualcomm case, the Guidelines also state it may be appropriate to use average avoidable costs (‘AAC’) or long-run average incremental costs (‘LRAIC’) as proxies for AVC and ATC respectively (paras 116-118, Qualcomm paras 435-443). When two-sided markets are concerned, the Guidelines suggest that it may be appropriate to include in the assessment revenues and costs incurred on both sides (para 120). The Guidelines use the hypothetical example of a market in which users are not charged and the platform is monetised by charging advertisers, where it may be appropriate to look at revenues and costs on all sides of the platform. The Guidelines do not suggest a separate test for platforms considering their low marginal costs like in this paper.
The test for margin squeeze
The concept of an equally efficient competitor features in this sub-section of the Guidelines (para 125). The Guidelines require the margin available to an equally efficient downstream competitor relying on the dominant undertaking’s input to be negative, assessed on the assumption that the competitor charges the same downstream price as the dominant undertaking. The Commission specifies that this is satisfied where the spread between downstream and upstream prices does not permit recovery of the incremental costs of supplying the downstream product, inclusive of the cost of capital (para 125). Where the spread is negative, no examination of downstream costs is needed at all, as it is clear that the equally efficient competitor will not be able to operate profitably (para 126).
The final text can be read as abandoning the 2024 Draft’s alternative formulation, under which a margin might be abusive where it failed to cover product-specific downstream costs without being outright negative. The adopted language is more equivocal than that reading allows, because paragraph 125 does retain an incremental cost benchmark. The emphasis has nonetheless moved towards the negative spread scenario in paragraph 126, and the practical question for advisers is how much room now remains to challenge thin but positive margins in vertically integrated industries.
Further, the Guidelines confirm that dominance is required only upstream and that the form of vertical integration is immaterial, a position reaffirmed in the December 2025 in Lukoil Bulgaria (para 124; Case C-260/24, Lukoil Bulgaria,paras 33, 35 and 41). Neither excessive upstream pricing nor predatory downstream pricing needs to be shown independently, and that recoupment is again beside the point (para 127). The Guidelines compress the theory of harm into a single move, reasoning that because a margin squeeze necessarily drives an equally efficient competitor into negative territory, exclusionary effects are probable rather than merely possible (para 128).
Rebates, exclusivity, and what survives of the presumptions
Rebates not conditional on exclusivity are handled under the general framework rather than through a dedicated test, with the Commission permitting reasoning by analogy from both the exclusivity rebate line and the predatory pricing line (paras 136-150). The relevant considerations include the extent of market power, the duration and coverage of the scheme, whether the dominant undertaking is an unavoidable trading partner, and the relationship between the rebate and the applicable threshold (para 142). The mechanism operates through the suction effect on the contestable share of demand, where a retroactive rebate concentrated near a threshold makes marginal units disproportionately expensive for a rival to win.
Exclusive dealing is the clearest surviving instance of presumptive treatment (paras 151-161). The Guidelines state that exclusive dealing is subject to a presumption that it distorts effective competition (para 155), extending the logic of Intel beyond exclusivity rebates to exclusivity obligations as such (Case 680/20, Unilever, para 51).
The Guidelines differentiate between de jure (an explicit arrangement) and de facto exclusivity which includes arrangements that are not explicit but produce equivalent commercial effects (paras 152-154). The Guidelines then identify the rebuttal evidence, mainly limited coverage and limited duration (para 157). Whether this counts as a meaningful softening relative to the 2024 Draft is contested, and the Kluwer commentary linked here characterises what survives as comparatively soft presumptions.
Tying, access restrictions and refusal to supply
The Guidelines restate the four cumulative conditions for abusive tying, requiring separate products, dominance in the tying market, coercion, and capability of producing exclusionary effects (para 168). Separateness turns on independent customer demand rather than substitutability, which keeps the inquiry distinct from market definition. The Guidelines confirm that coercion may exist even where the tied product carries no separate charge and even where the customer is never obliged to use it (paras 174-175). That is directly relevant to any analysis of default arrangements in mobile ecosystems, a point that connects closely to the reasoning discussed in our analysis of the Google Android judgment on this blog.
Access restrictions form the broadest of the newly systematised categories, covering denial of access on commercially viable conditions as well as conduct that hinders or delays it (paras 179-183). The Guidelines clarify that indispensability is not a precondition, functioning instead as a factor bearing on the likelihood of exclusionary effects (para 181). Refusal to supply is confined to inputs developed by the dominant undertaking solely for its own use, and the stricter treatment is justified on the ground that an access obligation impinges directly on freedom of contract and the right to property (para 184). The Guidelines preserve indispensability alongside the requirement that the refusal be capable of eliminating all effective competition for the requesting undertaking (paras 185-188), while also including the familiar requirement in intellectual property scenarios that the refusal obstruct a new product or otherwise limit technical development (para 189).
Self-preferencing, three factors in search of a mechanism
Self-preferencing receives its own subsection for the first time in Commission soft law, and opens cautiously by denying any general rule that the practice is problematic (para 191). The Guidelines supply the structure, describing the leveraging of dominance in one market to favour the undertaking’s own products in another, and distinguishing offensive leveraging aimed at strengthening position in the leveraged market from defensive leveraging aimed at protecting the dominated market (para 192).
The mechanism is articulated through the three factors (para 196). The first concerns whether the leveraging market is an important source of business that competitors in the leveraged market cannot effectively replace, a threshold deliberately pitched below Bronner indispensability, since the Commission confirms that the Bronner conditions have no application to self-preferencing at all (C-48/22 P, Google Shopping, paras 170-171). The second concerns unjustified difference in treatment, arising either where the conduct contradicts the undertaking’s own commercial rationale in the leveraging market or where an expectation of neutrality or openness has been created and then departed from in a non-objective and non-transparent manner (C-48/22 P, paras 146, 158 and 187). The third concerns influence on user behaviour irrespective of the intrinsic qualities of the favoured product, which the Commission ties explicitly to behavioural bias and to the reasoning of the Court of Justice in (C-612/17, Google Shopping, para 172; C-738/22 P, Google Android, paras 213 and 226).
Conduct that is harmful by its very nature
The Guidelines establish a ‘naked restrictions’ category by reference to conduct serving no economic interest for the dominant undertaking other than the restriction of competition, and therefore falling manifestly outside competition on the merits (para 197). Four illustrations are offered covering payments conditional upon customers declining or postponing a named rival’s products, coerced product swaps enforced through the threat of withdrawing discounts, the active dismantling of infrastructure on which a competitor depends, and discretionary gatekeeping by an undertaking holding regulatory and commercial functions simultaneously without transparent, objective and non-discriminatory criteria (para 198).
Having stated that such conduct is deemed as such to distort effective competition, the Guidelines acknowledge that the Union courts have not determined whether an undertaking may rebut the classification by showing that the conduct was incapable of producing exclusionary effects in the circumstances, and indicate that any such challenge could succeed only very exceptionally (paras 199-201). It then goes considerably further, stating that if the courts conclude, on a consistent reading of Articles 101 and 102, that the category should attract the evidentiary standard applicable to restrictions by object, the Commission will adapt its practice accordingly (para 200).
That invitation is difficult to reconcile with the case law, which is thinner than the passage implies. Assimilation would render actual and potential effects irrelevant to characterisation, and it would do so in a field where the Court has repeatedly insisted that Article 102 analysis remains effects-based. It is surprising for a soft law instrument to invite its own supersession in this way, and unusual again for the invitation to point towards a standard that would relieve the enforcer of proving effects at all.
Where the objective justifications sit
The Guidelines preserve the two established routes: objective necessity and efficiency, with the burden resting on the dominant undertaking throughout (para 202). The Guidelines list acceptable necessity grounds, including protection against unfair competition, protection of the integrity or security of a service, and the clearance of excess or obsolete stock through below-cost sales, while excluding mere price alignment with non-dominant rivals (para 208). A demanding position on interoperability is adopted, holding that the absence or difficulty of a technical solution will not suffice unless interoperability would compromise integrity or security, or is technically impossible (para 209).
The efficiency defence requires four cumulative conditions, namely that the conduct gives rise to efficiencies, that those efficiencies counteract the adverse effects on competition and consumers, that the conduct is indispensable to achieving them, and that effective competition is not eliminated (para 220). The sliding scale reappears in an unforgiving form, since the greater the conduct’s capacity to harm competition, the less likely the conditions are to be satisfied, which makes the defence close to unavailable for conduct falling within the by-nature-harmful category (para 222). The recognition of sustainability as a qualitative efficiency is the most novel element (para 218), although the requirement that the consumers harmed substantially overlap with those benefited constrains its practical reach considerably (paras 231-232).
A missed opportunity: exploitative abuse remains in oversight
Like the 2009 Guidance, the Guidelines also focus solely on exclusionary abuse. The Guidelines’ discussion of exploitative abuse remains limited to stating that the general principles governing dominance and objective justification also apply to other forms of abuse, including exploitative conduct (para 10). This choice leaves an important area of Article 102 TFEU enforcement outside the Guidelines’ framework. Exploitative theories of harm have become increasingly prominent in debates on the digital economy, and the Commission has also taken enforcement action in this area. The tendency to rely on exploitative theories of harm to address novel concerns in digital markets has made the need for clearer guidance more apparent, as the limited and fact-specific case law often leaves its application to new forms of conduct uncertain. Therefore, it would not be wrong to say that the choice to exclude exploitative abuses leaves the broader modernisation of Article 102 TFEU incomplete for now.
Where does this leave us?
Much has happened since the 2009 Guidance. Digital transformation has brought new challenges, novel forms of abuse have emerged, and established theories of harm have evolved, with a shift in the Commission’s policy towards a ‘workable effects-based approach’. These developments, alongside calls for greater legal certainty in abuse enforcement, have made updating the 2009 Guidance essential. The Guidelines will undoubtedly be at the forefront of the discussion as their influence unfolds. Yet, some aspects to look out for in future cases are: whether the theory of harm requirement acquires any disciplining force in practice; whether the Commission’s invitation regarding by-nature-harmful conduct is taken up by the Court, which would represent the most substantial doctrinal shift the document contemplates; how the retreat from the 2024 Draft’s presumption architecture actually plays out in non-pricing cases, where the Guidelines simultaneously narrow the formal presumptions and acknowledge that the equally efficient competitor concept may have no application in digital ecosystems at all. The Guidelines are more cautious than the 2024 Draft. However, the Commission, Courts, businesses, and other stakeholders now have a much-needed soft-law instrument that was long overdue.
