When Courts Meet Code: Judicial review of competition and DMA decisions

By Dr. Konstantina Bania

By the time a digital‑markets case reaches a court room, a paradoxical situation has arisen: the business model in the file is obsolete, the market definition section reads like historical fiction, and the junior in chambers is already using the platform that judges were told “doesn’t really exist”. That mismatch between legal time and digital time sits at the heart of the challenges facing courts when they review decisions affecting digital markets.

This post explores how these challenges play out both for classic competition enforcement and for DMA cases.

1. Why digital markets are different for courts

Digital‑market cases have a distinctive profile that makes them harder to review than traditional cartels or mergers for the reasons set out below.

First, competition authorities are pushed into quasi‑regulatory roles. They do not just impose fines; they scrutinise algorithms, and design and oversee complex conduct remedies.

Second, the evidential record looks very different from other cases. Instead of a finite bundle of documents and a few econometric reports, courts now see vast troves of emails and internal presentations, large datasets, simulations, and A/B tests.

Third, the remedies are often behavioral and forward‑looking. Such remedies impose interoperability mandates, data‑access obligations, UI changes, ranking tweaks, and functional separation. Courts are asked to assess not only whether the authority proved an infringement, but also whether the design of these remedies is proportionate and technically realistic.

All this plays out under intense time pressure and political salience. A “simple” manifest‑error check can feel too thin; but if courts go fully into the substance, they risk turning themselves into super‑regulators.

Against this backdrop, the central question becomes: how can courts ensure effective legal protection without paralysing adaptive enforcement in fast‑moving markets?

2. Challenges facing courts when reviewing competition decisions

Challenge 1: Standard and intensity of review

Broadly speaking, two models exist when it comes to the judicial review of (competition) decisions:

• Merits‑type review, where the court re‑examines facts, economics and discretionary assessments; and

• A more deferential judicial‑review model, closer to classic administrative law, where the focus is on legality, procedure and manifest error rather than on re‑deciding the case.

Digital cases strain this dichotomy. Even under a judicial‑review standard, courts increasingly face voluminous records mixing business, technical and experimental evidence; contested economic theories about network effects, multi‑sided markets and data advantages; and the risk that important mistakes of fact are buried in opaque technical materials. In practice, courts are no longer just arbitrating between two legal submissions; they are arbitrating between competing economic and technical world‑views about how a platform works. That is a qualitatively different task from testing, say, the definition of a cement market.

Challenge 2: Complex economic and technical evidence

Digital‑markets enforcement has shifted the substantive centre of gravity of many cases. Authorities now build theories of harm around API access, algorithmic ranking and self‑preferencing, and long‑run innovation effects (rather than short‑term price changes). For courts, this means learning to read and evaluate data‑science style outputs (e.g., funnel analyses, user‑journey evidence), descriptions of code, system architecture and compatibility constraints, and detailed discussions of how remedies would work “under the hood”.

A good illustration is the CJEU’s Android Auto interoperability ruling, where the Court revisited the classic Bronner indispensability criteria in the context of platform APIs (I’ve co-authored a blog about the ruling, which is available here). The Court had to decide under what conditions a platform must open interfaces to third‑party apps, while grappling with specific questions of architecture and security. Cases like Android Auto show that meaningful judicial review now requires a certain level of technical literacy – or at least a structured way to bring expertise into the courtroom.

Challenge 3: Predictive, forward‑looking analysis

The enforcement of competition law in digital markets is often predictive. Authorities act in markets that are still evolving and innovation‑driven, characterised by rapid entry and exit, and prone to tipping and entrenchment because of indirect network effects. The theories of harm are framed in terms of risk: the risk that conduct will tip a market, exclude innovative rivals, or entrench data advantages that cannot easily be undone later.

For courts, this raises hard questions:

• How much prediction is acceptable before a decision becomes speculative?
• When should a court say that an authority has gone beyond its margin of appreciation and no longer relies on verifiable evidence?

These tensions are particularly visible in merger control (e.g., “killer acquisition” scenarios) but also in cases affecting nascent markets where the counterfactual is inherently uncertain.

Challenge 4: Novel and evolving remedies

In classic competition cases, remedies were often one‑off and familiar: cease‑and‑desist orders, structural divestments, straightforward commitments. Digital‑markets cases look different. Authorities now reach for a diverse toolbox that includes data‑silo requirements, interoperability obligations, and changes to default settings, user interfaces, and recommender‑systems.

Courts reviewing these remedies must ask at least three questions:

• Proportionality: is the remedy truly tailored to the infringement, or does it effectively redesign the product or business model?
• Feasibility: is the remedy technically and operationally implementable, or is it an elegant theory that cannot work in practice?
• Dynamics: if the remedy is iterative and evolves over time, what exactly is the court reviewing: the decision as written, or the remedy as it has morphed by the time of judgment?

In other words, judicial review in this context often targets a moving object. That is a poor fit for traditional, snapshot‑in‑time judicial techniques.

Challenge 5: Intersections with other legal regimes

Finally, digital-markets cases rarely live in a silo. They intersect with data protection regulation (GDPR, ePrivacy); consumer protection rules; sector‑specific regimes (e.g., telecoms, media, financial services); and new digital‑regulation frameworks such as the DMA and DSA. This gives rise to several risks. Consider, for instance, a remedy requiring interoperability or data sharing that conflicts with privacy or security obligations, or an abuse finding grounded in “dark patterns” that sits uneasily with consumer and DSA case law. I discuss these tensions extensively here.

Therefore, courts have to reason horizontally across different legal regimes, not just vertically within competition law. They need to avoid contradicting other regulators, while keeping competition law within its proper bounds.

What can courts do? Towards a “digital judicial toolkit”

The picture so far may sound discouraging: more technical complexity, more prediction, more institutional tension. In practice, courts across Europe are already experimenting with ways to cope.

Four strategies stand out:

• Enhanced reasoning requirements: Courts can insist that competition authorities clearly articulate their theory of harm, counterfactuals and remedy design, so review is not reduced to rubber‑stamping.
• Use of expert knowledge: Judicial reliance on economic and technical experts – whether court‑appointed, amicus curiae, or via specialised tribunals – can make complex records digestible without requiring every judge to become a data scientist.
• Structured proportionality analysis: Making proportionality explicit (suitability, necessity, balancing) helps courts evaluate far‑reaching remedies in a disciplined way, especially where interventions re‑shape product design.
• Dialogue with other regimes: Conscious alignment with GDPR, consumer law, sectoral rules and, now, DMA/DSA case law helps ensure consistency and reduces the risk of contradictory obligations.

Taken together, these elements amount to a nascent “digital judicial toolkit”: this is not a substantive rewrite of competition law, but a set of procedural and analytical techniques tailored to platform cases.

3. Challenges facing courts when reviewing DMA decisions

Enter the DMA: a new kind of review

The same themes reappear – in sharper form – under the DMA regime. The DMA sits between classic competition law and sector regulation. Though it borrows concepts from competition law like market (read: gatekeeping) power and behavioural obligations, it relies on ex ante, per se style rules and tight deadlines, rather than open‑ended effects analysis.

For the EU Courts, early DMA appeals raise three overarching questions:

• How intensively to scrutinise the Commission’s economic and technical assessments, given the regime’s emphasis on speed.
• How to interpret novel legal concepts with little direct precedent.
• How to maintain coherence between DMA case law and classic Article 102 case law, especially where concepts overlap but legal tests differ.

These questions surface differently in designation (e.g., “who” is a gatekeeper, for “which” core platform services) and non‑compliance decisions (e.g., whether real‑world implementations of obligations are sufficient).

Challenge 1: Designation under the DMA is where the hard questions start

The designation stage is the entry gate to the DMA. Once a service qualifies as a “core platform service” and its provider is designated as a “gatekeeper”, the whole architecture of obligations, compliance reports and non‑compliance proceedings follows. That makes designation decisions the first real stress‑test for the EU Courts.

Three aspects are worth highlighting.

a. Defining “core platform services” and “gateways”

Many appeals will turn not on grand constitutional issues, but on apparently technical questions of scope:

• When is a product a single “core platform service”, and when is it a bundle of distinct services that should be assessed separately?
• How should hybrid models be treated – for example, platforms that mix marketplace, own‑retail and logistics, or integrate messaging into a wider social‑network ecosystem?
• What does it really mean to provide an “important gateway” in environments where users multi‑home across several apps and switch frequently?

These questions are not merely taxonomic. They shape how user numbers are calculated and, by extension, whether the DMA thresholds for designation are met; which parts of an ecosystem will later be subject to the DMA obligations; and how far the DMA can reach into evolving digital architectures without sacrificing legal certainty. Courts will be asked to strike a balance between flexible, future‑proof categories and predictable, justiciable boundaries.

b. Rebuttals and the strength of the presumption that a platform is a “gatekeeper”

The DMA relies heavily on quantitative thresholds (turnover and active‑user numbers) that create a rebuttable presumption of gatekeeper status. Designation appeals are already testing what “rebuttable” really means in practice.

Key issues for the Courts include:

• How strong is the presumption once the thresholds are met – is it merely a starting point, or something closer to a strong default that will only rarely be overturned?
• What kind of qualitative evidence is needed to rebut it – for instance, can a platform argue that rapid growth, intense multi‑homing or nascent competitive constraints show its position is not yet “entrenched and durable”?
• How wide is the Commission’s margin of discretion when assessing rebuttal arguments and mixed quantitative-qualitative evidence?

The early TikTok/ByteDance designation judgment (the summary is available here) suggests the EU Courts are inclined to treat the presumption as robust: even a relatively young service can be considered “entrenched” where there is sustained large‑scale use and clear network effects, and generic appeals to dynamic competition or multi‑homing are unlikely to suffice. That sets a high bar for companies hoping to escape designation once they meet the thresholds.

Challenge 2: Reviewing DMA non‑compliance decisions means shifting away from “bright‑line” rules to messy systems

If designation decisions decide who is in the DMA’s net, non‑compliance decisions decide what those gatekeepers must actually do. They are also where the promise of “bright‑line, ex ante rules” collides with the complexity of real‑world systems.

On paper, many DMA obligations look simple: bans on self‑preferencing, restrictions on combining personal data, duties to open up app‑stores or messaging services, non‑discrimination in access and ranking. In practice, enforcing those rules requires the Commission to evaluate detailed design choices in ranking algorithms, default settings, data‑sharing interfaces and APIs.

For the EU Courts, this raises a distinct set of review challenges.

1. Interpreting the scope of individual obligations

Non‑compliance appeals are expected to turn on how specific obligations are interpreted and applied to complex ecosystems. Courts will have to answer questions such as:

• What exactly counts as “self‑preferencing” in rankings, design and access conditions?
• To what extent can a gatekeeper invoke security, privacy or integrity justifications to limit interoperability or data access?

The DMA was deliberately drafted to be narrow and effects‑light. Courts will need to distinguish between genuine proportionality arguments and attempts to re‑import open‑ended balancing that would dilute the regime’s ex ante character.

In a non‑compliance decision, the Commission will often argue that a particular default setting, screen flow or ranking rule is inconsistent with an obligation; and an alternative design would restore compliance without undermining the core service. Courts must then decide whether the Commission has properly identified the line between what the DMA prohibits and what it allows; and how far they should second‑guess the Commission’s view on technical alternatives and implementation choices.

If they review too lightly, almost any intrusive design prescription may be upheld. If they review too intensely, they risk re‑writing the regime into something closer to traditional, effects‑driven antitrust.

• Horizontal coordination in non‑compliance settings

Non‑compliance decisions are where the DMA’s overlaps with competition law, data protection regulation and consumer protection rules become most acute. Data‑combination, tracking and interoperability cases will often sit at the intersection of DMA obligations, GDPR/ePrivacy constraints, and broader consumer‑ and platform‑regulation rules.

When reviewing non‑compliance decisions in such contexts, courts will be asked to:

• ensure that DMA‑driven design changes do not destabilise settled privacy or consumer‑protection doctrine;
• prevent gatekeepers from using privacy or security as pretexts to block competition‑enhancing changes;
• align reasoning across regimes so that firms are not pulled in incompatible directions by different parts of EU law.

This requires the courts to think explicitly in terms of horizontal coherence across EU digital regulation, rather than treating the DMA as a self‑contained code.

Put simply, designation appeals will define who is subject to the DMA and on what basis. Non‑compliance appeals will determine whether the regime can translate bright‑line obligations into workable, legally controllable design choices in real systems. The way courts calibrate their standard of review in those non‑compliance cases will go a long way towards deciding whether the DMA remains a fast, rule‑based instrument or slowly drifts back towards classic, effects‑based antitrust.

The stakes: legitimacy and adaptability

Stepping back, if courts are too deferential, there is a risk of unchecked regulatory power and error‑prone experimentation in markets where remedies can reshape global ecosystems. If they are too intrusive, enforcement may freeze or become so slow and cautious that it cannot address problems in time.

Finding the right balance will determine not only individual outcomes, but also the long‑term legitimacy and effectiveness of Europe’s digital‑enforcement architecture. In that sense, the real “platform battles” of the coming years will be fought as much in the courtroom as in the codebase.

The image was AI generated.