ICLE Comments to the CMA on Proposed Steering Conduct Requirements for Apple and Google
Introduction and Summary
The International Center for Law & Economics (ICLE) welcomes the opportunity to respond to the Competition and Markets Authority’s (CMA) consultations on proposed steering conduct requirements (CRs) for Apple and Google.[1] ICLE is a non-profit, non-partisan global research and policy centre that advances evidence-based policy. Its scholars have written extensively on competition in digital markets. Because the two draft CRs are materially identical, these comments address both consultations together.[2]
Two points frame our response. First, the CMA’s proposed intervention rests on designation findings that ICLE has contested in earlier submissions. We do not repeat those arguments here, but the premise remains disputed. The extent of competition between iOS and Android bears directly on both the likely benefits of intervention and the costs of regulatory error.[3]
Second, these comments build on ICLE’s April 2026 response to the CMA’s call for evidence, which examined the fee, design, and security questions the draft CR now seeks to resolve.[4] Where appropriate, we cross-refer to that submission rather than repeat its analysis.
One distinction organises everything that follows: the draft CR combines two fundamentally different interventions.
The first is a steering-rights remedy. Paragraphs 3 to 8 and 14 would remove restrictions on communicating with end users and linking to external transactions. They would also require equal treatment of redirection mechanisms, permit a single neutral interstitial screen, and prohibit discrimination against developers that steer.
The second is a price-control regime. Paragraphs 9 to 13 would cap any steering fee at ‘no higher than’ the output of a forward-looking long-run incremental cost (LRIC) model plus a ‘value’ assessment adjusted to remove the effects of the market position that prompted designation. The regime would be enforced through cost-accounting obligations, a requirement that each firm demonstrate compliance ‘to the satisfaction of the CMA’, and quarterly reporting.[5]
These components require different responses. The steering-rights core can be designed proportionately, and much of it resembles measures already taking effect worldwide—including in the United Kingdom, where Google introduced steering on 30 June 2026.[6] Our concerns are principally matters of calibration. The ‘strictly necessary’ standard for platform safeguards is too restrictive, while the interstitial-screen provisions risk suppressing truthful information that consumers may need to make informed decisions. Section IV develops these points.
The fee provisions are different in kind. They would graft the machinery of utility-rate regulation onto an information-goods platform, relying on pricing principles that cannot produce stable or economically coherent results. No other jurisdiction—not the U.S. courts in Epic Games v. Apple, Japan under the Mobile Software Competition Act, or even the European Union under the Digital Markets Act (DMA)—has adopted anything comparable. Sections II and III explain why the proposed principles are indeterminate where they are defined, undefined where they matter most, and structured to operate as a one-way ratchet towards incremental cost.
The CMA’s provisional proportionality assessment does not cure these defects; it assumes them away. It discounts the platforms’ lost revenue through circular reasoning, treats near-complete pass-through of savings as a given, dismisses ecosystem-wide costs on the ground that firms may simply decline to steer, measures benefits against a no-steering counterfactual that no longer exists, and declares the proposal ‘the least onerous’ option without assessing an obvious graduated alternative. Section IV addresses each flaw.
Section v turns to the dynamic consequences. Compressing returns towards incremental cost would misprice the innovation that has generated much of the consumer welfare associated with these ecosystems.
Section VI sets out our recommendations. In brief, the CMA should separate the two components. It should proceed with an amended steering-rights CR but replace paragraphs 9 to 13 with a ‘fair and reasonable’ obligation assessed ex post against commercial benchmarks. Transparency requirements should support that obligation, together with a defined trigger for further intervention if evidence shows that fee levels have frustrated the remedy.
The stakes extend beyond steering. These proposals are among the first substantive CRs under the Digital Markets, Competition and Consumers Act 2024 (DMCC), and they will help set the pattern for those that follow. Parliament and the CMA presented the DMCC as a flexible, participative, and proportionate regime—a deliberate contrast with the DMA’s prescriptive model.[7] Consultations such as this one will determine whether that promise survives first contact with regulation.
I. The CR Conflates Steering Rights with Price Regulation
Paragraphs 9 to 13 are presented as part of the steering remedy, but they are an instrument of a different kind. A steering right removes obstacles between developers and alternative transaction channels, then leaves competition to determine the resulting terms. The fee provisions instead determine those terms administratively.
Section I.A shows that the provisions use the machinery of utility-rate regulation: LRIC modelling, common-cost allocation, regulated returns, and continuing supervisory oversight. Section I.B explains why the CMA’s frustration rationale supports strong non-price safeguards but not administered fees. On the consultation’s own theory, price regulation is either redundant because steering disciplines fees or a substitute for the competition the remedy is meant to create.
Section I.C turns to the institutional stakes. As one of the first substantive CRs under the DMCC, this measure will shape later interventions. Beginning with principles-based price administration would push the regime towards market ordering rather than market oversight. As Section VII explains, the CMA should instead reserve intervention on fees for evidence that pricing has frustrated an otherwise effective steering remedy.
A. The CR Imports Utility-Rate Regulation
The effect of the CR on steering fees becomes clearest in its mechanics. Those mechanics therefore warrant close attention.
Paragraph 10 provides that any steering fee must be ‘no higher than’ a fee calculated under specified pricing principles. The cost-based principle limits the fee to the costs of services ‘used by a Developer to a material and direct extent in facilitating or supporting a Steered Transaction’. Those costs must be calculated using ‘a forward-looking long-run incremental cost approach’, with ‘an appropriate recovery of common costs’ and ‘a reasonable rate of return as appropriate’.[8]
The CMA’s own explanation is revealing. It acknowledges that the LRIC of serving steered transactions ‘could be very low’ because most app-store operating costs would arise regardless of whether an app steers users elsewhere. On that view, the genuinely incremental cost ‘may be limited to the provision of technology that allows a developer to offer external links’.[9]
Paragraph 12 narrows the recoverable cost base further. It excludes ‘general app discovery, marketing, developer tooling’, as well as in-app-payment security and privacy features ‘not required for Steered Transactions’.[10] The value-based principle then requires the fee to be adjusted to remove the effects of the platform’s ‘substantial and entrenched market power’ and to account for the value that developers contribute across the platform.[11]
The CR backs these substantive constraints with extensive administrative machinery. Paragraph 9(a) places the burden on the firm to demonstrate compliance ‘to the satisfaction of the CMA’. Paragraph 12(d) requires cost-accounting records maintained for that purpose. The CMA also proposes rolling quarterly compliance reports tracking, among other things, ‘total platform charges applied to purchases’.[12]
This is the toolkit of utility-rate regulation—LRIC modelling, common-cost allocation, a regulated rate of return, and continuing administrative supervision of prices—applied to a hyperlink. The consultation makes the lineage explicit by comparing its cost-based methodology with the price controls imposed by the Civil Aviation Authority on Heathrow, Ofcom’s telecoms market reviews, Ofgem’s RIIO framework, and the Office of Rail and Road’s regulation of Network Rail access charges.[13]
That comparison exposes the problem. Those regimes govern bottleneck infrastructure whose owners cannot practically prevent incremental use. Here, the CMA would regulate what a firm may charge developers to route transactions away from its own payment system.
The consultation does not hide the intended direction of travel. It ‘expect[s]’ fees to fall under the CR and treats that result as evidence of success.[14]
B. Preventing Frustration Does Not Require Price Control
The consultation’s central justification deserves a fair hearing. The CMA argues that evidence from its investigation and ‘experience in other jurisdictions’ shows that merely prohibiting restrictions on steering would not suffice. Platforms would retain both the ability and incentive to impose ‘restrictive conditions, complex design requirements, or commercially unattractive terms’ that frustrate the remedy.[15]
The concern is legitimate, and the history of comparable interventions supports it. A bare prohibition on anti-steering rules may invite platforms to recreate the same restriction under another name. The question is what follows from that concern—and, specifically, whether it justifies moving from removing restrictions to administering prices.
The analysis must distinguish between two ways a remedy might be frustrated. A platform might use non-price conduct to recreate friction, or it might charge a fee that makes steering commercially unattractive. Those risks call for different responses. Conflating them allows the consultation to present rate regulation as the natural completion of a steering right.
The draft CR already addresses non-price frustration without regulating prices. Paragraph 4 extends the steering-rights obligations to any action ‘whether direct or indirect, and irrespective of the means by which it is implemented, including contractual terms, technical design requirements, app review processes, enforcement practices or other measures’. Paragraph 14 separately prohibits discrimination against developers that steer.[16] Together, those provisions reach precisely the conduct the CMA identifies.
Pricing terms are different in kind. On this point, the fee provisions invert the CMA’s own theory of the case. The consultation treats steering as a source of competitive discipline. Giving developers access to ‘other, lower cost distribution channels’ should exert ‘downward pressure on wider App Store commission fees’. The CMA likewise identifies the insulation of platform fees from competitive pressure as the problem the remedy should correct.[17]
Under that theory, the fee level is not an input into the remedy. It is the market outcome that an effective steering regime should discipline. Paragraphs 9 to 13 instead determine that outcome administratively.
This leaves a dilemma the consultation does not confront. If steering operates as the competitive constraint the CMA anticipates, the resulting downward pressure on fees makes rate regulation redundant. If steering does not provide that constraint, capping the fee near incremental cost does not create competition; it substitutes an administered price for a market price. In neither case is the price control merely ancillary to the steering right. It displaces the competitive process the remedy is supposed to restore.
The claim that pricing latitude will nullify the remedy is empirical and should be tested. The consultation itself supplies relevant comparisons. Steered-transaction fees of 10 to 20 per cent have coexisted with commercial uptake in Japan and through Google’s global rollout. Google’s UK measures, introduced on 30 June 2026, will soon provide direct domestic evidence.[18]
The CR could reserve further intervention for evidence that fee levels have frustrated steering, as our recommendations propose, rather than presume frustration and regulate prices from the outset.
The CMA’s proposed effectiveness metric underscores the problem. It plans to monitor ‘developer and user uptake of steering’ as a measure of success,[19] effectively treating steering volume as an end in itself. Yet the statutory objectives of fair dealing and open choices concern the terms on which developers and users may transact, not the market share of any particular payment channel.
A developer who receives a genuine, unrestricted choice and elects to retain in-app purchase has not been failed by the remedy. Under the consultation’s metric, however, that choice risks appearing as a regulatory shortfall that demands correction.
C. The CR Risks Turning Oversight into Market Ordering
The distinction between the CR’s two components reflects a broader institutional divide. Market-oversight regimes police the competitive process through effects analysis, efficiency defences, and meaningful review. Market-ordering regimes instead design outcomes directly through ex ante obligations, price regulation, and continuous administrative supervision.[20]
This is more than a question of labels. It concerns who decides the terms of trade. Under oversight, an authority asks whether conduct has harmed competition and, if so, acts to stop it. Under ordering, the authority determines what the terms of dealing should be and supervises the firm’s compliance over time.
The DMA has moved decisively towards market ordering. ICLE has documented the predictable consequences: indeterminate standards interpreted by the enforcer, prolonged compliance disputes, and an institutional structure poorly equipped to identify and correct its own mistakes.[21] The DMCC was designed—and presented—as something different: a participative regime whose interventions would be targeted, proportionate, and evidence-led.
The European Commission’s proceedings over Apple’s fees offer a preview of where paragraphs 9 to 13 may lead. After challenging Apple’s 30 per cent commission and imposing a €500 million fine, the Commission subjected each successive revision to Apple’s fee structure—the Core Technology Fee, the initial acquisition fee, and the tiered Store Services fee—to another round of scrutiny. Years of enforcement have yet to produce a stable outcome for developers, consumers, or the platform.[22]
The draft CR recreates the conditions that generated those disputes: open-textured standards such as ‘fair and reasonable’, ‘material and direct extent’, ‘appropriate recovery’, and an adjustment for ‘market power’; a firm-side burden of proof; and rolling compliance adjudication. It adds one significant aggravating factor: decisions under the DMCC are reviewable only on judicial-review principles, not on the merits.[23]
That limited review strengthens the case for restraint before the CMA reaches for price regulation. A regulator whose pricing judgement cannot readily be corrected on appeal should be correspondingly reluctant to set the price in the first place.
The design of this CR will also travel. As one of the first substantive CRs under the DMCC, it will become a template and reference point for later interventions involving these firms and others. Future requirements will be drafted, negotiated, and defended in its shadow. A precedent for principles-based price-setting established here will not remain neatly confined to steering fees.
Beginning the regime with the most prescriptive attempt at platform-price regulation in any jurisdiction would place the DMCC on the market-ordering path from the outset. As ICLE explained in its comments on the CMA’s draft guidance, the sequencing should run in the opposite direction: begin with simpler interventions whose effects can be observed, then escalate as evidence and experience warrant.[24]
II. The Pricing Principles Cannot Produce a Coherent Fee
The CMA asks whether the proposed pricing principles are appropriate, how cost-plus and value should be defined, and how the two should interact.[25] They are not, for reasons that go to the structure of the framework rather than its wording.
The cost-based principle cannot allocate the joint costs and value of an information-goods platform without arbitrary regulatory judgements. The value-based principle requires a competitive counterfactual that the CMA cannot construct and rejects the benchmarks that might inform it. Applied together, the principles create a one-way ratchet towards incremental cost, reinforced by attribution rules that exclude much of the value platforms continue to provide. If the CMA nevertheless assesses fee levels, cautious benchmarking against comparable services offers the least-bad approach—and the available evidence does not support intervention.
A. Cost-Based Pricing Cannot Capture Platform Value
App stores are paradigmatic information goods. They require high fixed and sunk costs, while the marginal cost of distribution approaches zero. They also rely on shared inputs—security infrastructure, review and curation, developer tools, and anti-fraud systems—that support millions of downstream services and resist clean allocation among them.[26]
As ICLE has previously explained, cost-based rate-setting in this setting presents three difficult choices. A regulator can allocate shared fixed costs across services, confine recovery to incremental costs, or add a return on capital to a selected asset base.[27] Paragraph 10(a) adopts all three at once: LRIC, ‘appropriate’ common-cost recovery, and a ‘reasonable’ rate of return. That combination compounds the indeterminacy rather than resolving it.
Allocating common costs requires assumptions about how to divide expenses among interdependent products and services. No objective economic rule supplies the answer. Restricting recovery to incremental costs avoids that allocation problem, but risks excluding the platform-wide investments needed to create, maintain, and improve the service. Adding a return on capital may appear to correct that undercompensation, but it introduces new disputes over the asset base, the appropriate rate of return, and the period over which investment should be recovered.
Cost-based analysis therefore does not eliminate regulatory judgement. It merely relocates that judgement into the construction of the benchmark.
Paragraph 12 makes the problem concrete. It excludes from the cost base the platform-wide investments that give a steered transaction much of its value, including discovery, marketing, developer tools, and payment-security features ‘not required’ for the steered sale. The apparent logic is that a developer who steers does not use those services to a ‘material and direct extent’ and therefore should not pay for them.
That misunderstands what the developer receives. Steering removes one step from the transaction: payment processing. The developer continues to benefit from the rest of the platform bundle, including worldwide distribution and updates, discovery and merchandising, developer tools and APIs, and the consumer trust created by curation and fraud prevention. The commission was never merely a payment-processing fee. It was the price of that broader bundle, of which payment processing forms only one part.
A steered transaction therefore monetises a customer relationship that the platform helped create and continues to support. ‘Direct use’ accounting cannot divide value that is jointly produced. Paragraph 12 captures only the platform’s narrow co-ordination function at the point of linking a developer and user, while ignoring the wider value of the platform that made the transaction possible.
The exclusions also sit uneasily with the CMA’s own findings. The SMS decisions define the relevant digital activity as an integrated ‘Mobile Platform’ comprising the operating system, native app distribution, and the browser.[28] Paragraph 13(e) of the draft CR likewise recognises value flowing from ‘all elements’ of that platform. Yet the cost-based principle limits recoverable costs to a narrow slice of native app distribution. The result is an asymmetry between the value recognised and the remuneration permitted.
A second asymmetry runs through the cost-plus framework. App-store economics are a joint enterprise. The platform invests in the operating system, distribution, tools, and security. Developers invest in their apps. Together, they create a surplus that must be divided.[29]
How that surplus should be split—whether the platform’s share is 30 per cent, 15 per cent, or some lower figure—is a legitimate question, and benchmarking may help answer it. Paragraph 10 does not ask that question. It examines only the platform’s costs and only the platform’s return, as though the platform were a stand-alone utility rather than one participant in a two-sided exchange. It does not account for developers’ costs, risks, or returns.
A method that addresses surplus division by measuring one side’s costs and capping one side’s return is not a neutral measure of fairness. It decides in advance that the platform’s share should tend towards incremental cost, leaving the residual to the other side.
This is familiar territory. The 9th U.S. Circuit Court of Appeals’ recent suggestion in Epic that a permissible commission might be limited to costs ‘genuinely and reasonably necessary for [Apple’s] coordination of external links’ imports the same mistake.[30] ICLE’s amicus filings explain the consequences. Courts and agencies that undertake to supervise the price of platform access assume a task the U.S. Supreme Court has repeatedly warned they are ill-suited to perform. The district court’s interim answer—a mandated price of zero—illustrates the arbitrariness that follows.[31]
Antitrust doctrine recognises why returns above incremental cost may be necessary: they finance the fixed-cost creation on which innovation depends. That is also a central premise of intellectual-property law.[32]
B. The Value Principle Supplies Discretion, Not a Benchmark
The value-based principle might appear to correct the cost principle’s myopia. As drafted, it cannot. Paragraph 13(d) requires the fee to ‘adjust for the direct and indirect effects of’ the platform’s ‘substantial and entrenched market power’ across all elements of the Mobile Platform. In practice, that requires the firm to calculate the price that would prevail in a competitive market that has never existed, without a prescribed methodology or any benchmark the CMA is prepared to accept. The consultation concedes that the exercise is ‘complex by its nature and open to a degree of subjectivity’.[33]
The CMA’s treatment of benchmarks proves the point. Google proposed using comparators to separate competitive value from any market-power premium. The CMA rejected them all, reasoning that an appropriate comparator must ‘(i) provide similar services to those being regulated, (ii) be exposed to effective competition, and (iii) reflect the two-way value exchange between the participants and the platform’.[34]
Taken seriously, those conditions exclude any observable benchmark. A platform similar enough to satisfy the first and third requirements will, on the CMA’s own premises, display the features that disqualify it under the second. A test that no evidence can satisfy is not a pricing principle. It is discretion.
The authority on which the CMA relies does not support the weight placed upon it. The Competition Appeal Tribunal’s observation in Kent v Apple that Apple’s headline commission appeared ‘arbitrary’ remains under appeal, arose in a damages action under a different legal standard, and ultimately cuts the other way. The Tribunal’s difficulty in identifying the fee that value would justify illustrates why administered value-pricing for information goods is so unstable.[35] A finding that Apple did not set its fee by reference to a demonstrable measure of value does not establish that a regulator can do so.
The obvious analogy is to the ‘fair and reasonable’ rates that courts set for standard-essential patents. But the FRAND analogy breaks down at its foundation. As ICLE has explained, FRAND obligations arise from an ex ante commitment made before lock-in and in exchange for the benefits of standardisation. That hypothetical pre-lock-in negotiation anchors every FRAND methodology.[36] There is no comparable bargain to reconstruct.
Even with such an anchor, FRAND adjudication produces ranges rather than determinate prices. The competing rates in Microsoft v Motorola differed by roughly thirty-fold, while the rate in Optis v Apple increased roughly sevenfold on appeal.[37] Those divergences arise because outcomes depend on methodological choices for which no canonical answer exists: which licences are comparable, how lump-sum payments should be converted into per-unit rates, how real-world agreements should be adjusted for hold-up or hold-out, whether to cross-check against the aggregate royalty burden, and which royalty base to use. Courts disagree even about the underlying conceptual framework.
Paragraph 13(e) compounds the problem by requiring the fee to deduct ‘the value contributed by Developers (in aggregate)’. That instruction misreads two-sided economics. Indirect network effects—the value each side receives from the other’s participation—are not a deduction from the platform’s contribution. They are the product the platform creates by bringing both sides together.
Platform prices therefore balance the two sides jointly. Effects must be assessed across the market as a whole, and a price observed on one side of a two-sided platform reveals little in isolation. Treating developer participation as a subtraction from platform value mistakes the mechanism of value creation for a reason to deny remuneration for it.[38]
C. The Combined Principles Create a One-Way Ratchet
Individually, the pricing principles are indeterminate. Together, they point in only one direction.
Paragraph 10 requires the fee to be ‘no higher than’ a level consistent with the principles, while the consultation says that cost and value should serve as ‘a cross-check on each other’.[39] But because the firm must demonstrate compliance ‘to the satisfaction of the CMA’, that cross-check can only reduce the defensible fee, not increase it. Value evidence cannot justify a fee above the cost model’s output, while a low cost estimate caps the fee regardless of what the value analysis shows.
The promised ‘degree of flexibility’ is therefore flexibility in one direction. The third principle—administrative simplicity—only sharpens the irony. The CMA notes that current UK fee structures are ‘relatively simple’ and that arrangements elsewhere have become ‘overly complex’.[40] Yet the CR would import the very machinery that produced that complexity: cost models, attribution accounting, monitoring APIs, and continuing supervisory review.
Paragraph 9(b) hard-codes commercial terms with the same downward bias. A steering fee may apply only when the transaction occurs within seven days of redirection, and no fee may attach to a subscription that ‘renews automatically and was entered into prior to’ the redirection. Together, those provisions exclude much of the lifetime value of a steered customer relationship from compensation. The under-remuneration is not incidental. It is built into the design.
The seven-day cut-off also rests on a mistaken account of how platform value is created. It assumes a discrete referral event, as though the platform’s contribution ended once the user reached an external page. In reality, platforms continue to support the relationship through updates, notifications, discovery, and re-engagement tools that keep users active long after any single redirection. A customer is retained over time, not simply ‘acquired’ once.
Attribution windows are also ordinary commercial terms, common in affiliate and referral arrangements. As ICLE has explained, they do not admit of precise regulatory calibration, and no economic principle makes seven days—or any other fixed period—the uniquely correct answer.[41]
D. Comparable Fees Do Not Support Intervention
If the CMA nevertheless insists on assessing fee levels, benchmarking comparable services remains the least-bad method. Applied conservatively, it does not support intervention.
Gaming-console storefronts charge 30 per cent. Steam charges 30 per cent, falling to 20–25 per cent at higher volumes. Amazon Marketplace referral fees range from 8 to 45 per cent, while food-delivery and online-travel platforms commonly charge 15–30 per cent. Against that range, steered-transaction fees already in force—15 and 10 per cent in Japan, 10–20 per cent in Google’s global rollout, and the EU’s alternative business terms—are unremarkable and often lower.[42]
Developer dissatisfaction does not change the analysis. Firms naturally prefer lower input prices, but that preference does not show that prevailing rates depart from competitive norms. The CMA’s legitimate concern that fees remain ‘commercially viable for developers’ is empirical.[43] Observed adoption provides the relevant evidence: where steering has launched at fees within the benchmark range, developers have used it. Google’s UK rollout should provide a domestic data point within the year.
At the other extreme, some developers argue that any fee on a steered transaction is illegitimate because the sale occurs ‘outside’ the platform. The U.S. district court’s zero-commission order embraced that position, and its defects are instructive.
A steered sale is not independent of the platform. The developer acquired, retained, and re-engaged the customer through infrastructure it did not build. The ability to prevent free riding is what makes investment in that infrastructure worthwhile. Anti-steering rules and access fees are alternative means of serving the same function. Removing the first while prohibiting the second would turn platform access into a mandatory gift.
Amex recognised that vertical restraints of this kind may be procompetitive because they prevent free riding. A CR that permitted no meaningful fee would therefore leave the platforms with less protection than ordinary competition law provides.[44]
III. The CR Fails the DMCC’s Proportionality Test
Section 19(5) of the DMCC makes proportionality the principal legal constraint on the design of CRs.[45] The CMA must do more than identify a legitimate objective. It must establish, on evidence rather than assumption, that the measure is necessary, that no equally effective but less onerous alternative exists, and that its expected benefits justify its disadvantages.
The proposed CR does not appear to meet that standard. The assessment assumes away the loss of platform revenue, relies on unsupported claims about pass-through and innovation, overlooks system-wide costs borne beyond individual steered transactions, and never tests an obvious graduated alternative. It also lacks consumer-outcome metrics capable of revealing whether the intervention has failed. The result is not a proportionality analysis so much as a restatement of the case for intervention.
A. The CMA Assumes Away the CR’s Largest Cost
The CMA acknowledges that the CR will reduce platform revenue but declines to treat that reduction as a cost. It reasons that current fees reflect the platforms’ substantial and entrenched market power and their restrictions on steering.[46]
That conclusion is doing the work the fee inquiry was supposed to perform. Whether current fees exceed the level competition would produce is precisely the disputed question. The CMA concedes that it lacks a reliable benchmark for the competitive counterfactual, as Section III.B explains. The statutory framework nevertheless requires the CMA to assess the intervention’s disadvantages, not define them away. A transfer from platforms to developers counts as a pure social gain only if every pound transferred was improperly retained in the first place.
The consultation also claims that the regulated fee will ‘better reflect the price Apple/Google would be able to charge when faced with more effective competition’. Yet the effectiveness assessment acknowledges that the value adjustment needed to construct that price is ‘complex by its nature and open to a degree of subjectivity’. The CMA has also rejected every comparator against which the result might be tested.[47]
The proportionality analysis therefore excludes the CR’s largest cost by invoking a counterfactual price that the consultation elsewhere admits it cannot reliably construct.
Once the CMA assumes rather than demonstrates that current fees are excessive, the proportionality exercise no longer evaluates the intervention. It simply restates the premise of designation. That reverses the statutory order of analysis: the CMA must first identify and assess the disadvantages before deciding whether they are proportionate to the CR’s aims.
B. The Claimed Benefits Depend on Unsupported Pass-Through
The CMA’s headline benefit is that ‘tens of millions of pounds per year could be either retained by developers for further investments or passed on to UK end-users in the form of lower prices’.[48] Neither claim withstands scrutiny.
To the extent developers retain the savings, the CR produces a transfer. That transfer counts as a benefit only on the circular premise discussed in Section II.A—and it is not necessarily a consumer benefit, to which the DMCC requires the CMA to give particular regard.[49] The developers best placed to take advantage of low steering fees are large global firms, so much of the retained margin may accrue outside the United Kingdom.
To the extent developers pass on the savings, the projected consumer benefit depends on the rate of pass-through. The assessment assumes that lower steering fees will produce ‘modest price decreases’ and ‘millions of pounds’ in savings for UK consumers.[50] That result is far from self-evident, particularly in multisided markets.
Platforms optimise prices across distinct user groups. A constraint on one margin may therefore prompt rebalancing through quality, investment, ancillary charges, or prices elsewhere in the ecosystem. The incidence of a fee cap cannot be inferred from the reduction in the capped fee alone.[51]
Experience with other platforms illustrates the risk. Caps on payment-card interchange fees produced only partial reductions in consumer prices, while banks recovered lost revenue through higher account charges.[52] When U.S. cities capped delivery-platform commissions, platforms raised consumer delivery fees and shifted promotions towards chain restaurants outside the caps. Independent restaurants—the intended beneficiaries—received fewer orders and lower revenue.[53]
Distribution matters as much as aggregate magnitude. Large developers with recognised brands, in-house payment systems, and direct customer relationships are best positioned to exploit low steering fees. Smaller developers gain less and face greater exposure to rebalancing through higher charges elsewhere, reduced investment in shared tools, or diminished free distribution. As the CMA has acknowledged, smaller developers ‘often do not pay the highest commission rates’ and represent a substantial share of the UK developer base.[54]
The consultations’ case studies show who might gain but do not answer the incidence question. The Apple Consultation features a globally distributed social-media platform paying the 30 per cent headline rate. The Google Consultation considers a developer earning more than $1 million annually on the standard tier.[55] Neither examines how the proposal would affect the far larger population of small developers.
The claimed innovation benefits are also overstated. The prospect of ‘new products, new entry, [and] new opportunities to offer more seamless services’ flows from the steering right: the ability to communicate with customers, choose a payment processor, and structure offers independently.[56] Paragraphs 3 to 8 secure those gains. Paragraphs 9 to 13 do not.
Because the two components are severable, each must carry its own proportionality justification. The CMA cannot transfer the innovation benefits produced by the steering right to the separate price-control regime. Where the ultimate incidence depends on rebalancing across many participants, proportionality requires evidence, not assumption.
C. The Opt-Out Theory Ignores System-Wide Costs
The CMA treats costs to developers and end users as negligible because ‘both groups have the option to avoid steering, incurring no associated costs if they choose to do so’.[57] That reasoning fails for costs imposed at the system level rather than on individual transactions.
Security and fraud risks from large-scale redirection, reduced investment in shared infrastructure as its funding base erodes, added interface complexity, and the administrative machinery required by paragraphs 9 to 13 affect developers and users whether or not they steer. Non-steering developers may bear higher costs or receive fewer shared services, while users who never follow an external link may still face weaker security or a degraded platform experience.
The consultation’s own analysis also undercuts its opt-out premise. It acknowledges that ‘end-users may not always be in a position to accurately judge the privacy and/or security risks associated with particular decisions’.[58] The CMA therefore cannot assume that individual choice eliminates costs arising from the system-wide effects of steering.
The assessment dismisses one such cost with particular haste. It describes Apple’s compliance costs—including ‘the need to gather cost accounting data as a basis for setting fees’—as ‘unlikely to be substantial’.[59] Yet the same document repeatedly notes that Apple ‘does not in the ordinary course systematically allocate costs to specific products or services’.[60]
Paragraph 12(d) would therefore require Apple to build and maintain a product-level cost-accounting system that does not now exist, alongside quarterly reporting and monitoring APIs.[61] The consultation offers no estimate of the cost of that apparatus. Unlike transaction-specific costs, it would arise regardless of how many developers or users choose to steer.
More broadly, the assessment reduces the costs of steering to payment charges. Those charges help finance a bundle of jointly supplied and jointly consumed services, including distribution, discovery, developer tools, and security. Every participant benefits from those services, not only developers that use in-app payment.
Because the platform finances these services across the ecosystem, the proportionality analysis cannot stop with developers that choose to steer. It must consider how lower platform revenue may affect incentives to maintain investments that benefit non-steering developers and end users alike.
D. The CMA Never Tests the Least-Onerous Alternative
The consultation concludes that, because any effective measure ‘would require the removal of these restrictions’, the proposal is ‘the only effective measure, which is therefore the least onerous’.[62] That does not follow.
For the fee provisions, the CMA says only that it has ‘not identified any other equally effective conduct requirement designs’.[63] But the duty to choose the least onerous among equally effective measures requires a comparative assessment. The CMA cannot satisfy that requirement by declining to identify alternatives. On that approach, the test would be met whenever the authority advanced a single option, effectively reading the least-onerous requirement out of the framework.
The omission is especially striking because the CR’s two components are severable and an obvious graduated alternative exists. The CMA could proceed now with steering rights, non-discrimination, and fee transparency, while reviewing fees ex post against commercial benchmarks and defining a trigger for escalation. It could then reserve rate regulation for evidence that fees have frustrated the remedy. The consultation never assesses that option, and an alternative does not become less effective merely because it goes unexamined.
A graduated approach would also follow the CMA’s own logic. In defending the principles-based form of the fee provisions, the consultation notes that, if Apple or Google ‘failed to comply effectively with this higher-level CR’, the CMA could impose more detailed requirements.[64] The CMA likewise holds broader app-distribution measures—including ‘directly constraining Apple’s commission fees’—in reserve, ‘in particular if a steering intervention does not have the intended effect’.[65]
That is the sensible sequence: observe first, then escalate on evidence. The consultation applies it to every instrument except the steering fee and never explains the exception.
The asymmetry of error costs makes that omission consequential.[66] A CR that proves too weak can be strengthened in light of evidence. The DMCC’s review, revocation, and compliance-reporting provisions exist for that purpose.[67] A rate-regulation regime, once embedded in cost models, accounting systems, and reliance interests, is much harder to unwind.
The proposal also lacks any credible mechanism for recognising failure. Every metric the CMA proposes to monitor—steering uptake, transaction shares, platform charges, and abandonment rates—measures activity rather than consumer welfare.[68] None directly tracks prices, output, quality, fraud, or effects on smaller developers. Under those metrics, the intervention cannot fail; it can only ‘need strengthening’.
The European Commission’s first review of the DMA illustrates the danger. It catalogues enforcement outputs as evidence of success while treating stakeholder-reported harms as consultation feedback rather than findings.[69]
The CMA should therefore commit in the final decision to reassess the CR after a defined period against published consumer-outcome metrics, including prices, output, fraud and complaint rates, and effects on smaller developers. The decision should also provide expressly for narrowing or revoking the CR if its disadvantages outweigh its benefits.
IV. The CR Unduly Restricts Safeguards and Disclosure
Our disagreement concerns where the CR draws the line, not whether it should draw one. Steering may promote competition, and platforms may invoke safeguards as a pretext for obstruction. But the draft responds by constraining legitimate protections too aggressively.
Paragraph 3(b) recognises fewer justifications than the consultation’s own evidence supports, applies a narrower standard than the DMA, and fails to reflect the CMA’s stated commitment to protect security and privacy. The consultation also discounts emerging evidence that steering moves transactions from integrated fraud-prevention and dispute-resolution systems into developer-specific environments of uneven quality.
The interstitial provisions compound the problem by restricting truthful disclosures about the protections users lose when they transact off-platform merely because that information may discourage conversion. A regime committed to trust, transparency, and open choices should enable informed decisions, not sanitise them.
The subsections below address the appropriate scope of platform safeguards, the still-developing evidence on security and fraud risks, and the information a neutral interstitial screen should convey.
A. The Safeguards Standard Is Too Narrow
Paragraph 3(b) permits a platform to restrict a redirection mechanism only where doing so is ‘strictly necessary for an objectively justifiable purpose of preventing malware, fraud, unlawful content or content harmful to children’.[70] That closed list omits other legitimate risks, including threats to user privacy, scams and subscription traps that fall short of fraud, inadequate payment processors, and harms to vulnerable adults.
Those omissions are difficult to reconcile with the consultation’s own record. The CMA recounts, without contradiction, that dynamically generated links are inherently ‘gameable’: they may resolve to compliant destinations during review and change afterwards. It also notes that link parameters may ‘be used to exfiltrate information about the user from the app’, with pixel trackers collecting that information and transmitting it to third parties without the user’s knowledge or recourse.[71] Paragraph 3(b) does not adequately address those privacy risks.
The standard is also stricter than the regimes the CMA elsewhere treats as reference points. Even the DMA permits gatekeepers to adopt measures that are ‘strictly necessary and proportionate’ to protect the integrity of the operating system across its Article 6 obligations.[72] The 9th U.S. Circuit Court of Appeals in Epic likewise accepted Apple’s security and privacy justifications as legitimate and non-pretextual.[73]
The CMA’s stated objective is to preserve ‘appropriate safeguards to protect legitimate platform interests’, including security and privacy.[74] The CR should reflect that objective. Paragraph 3(b) should adopt a ‘reasonable and proportionate’ standard, and its list of legitimate purposes should be non-exhaustive and expressly include privacy and consumer protection.
The consultation responds that steering risks ‘may be manageable’.[75] But that assumes platforms retain the tools needed to manage them. As drafted, paragraph 3(b) withdraws those tools ex ante.
B. Steering Shifts Transactions into Riskier Environments
The consultation records, but largely discounts, the platforms’ evidence about what steering bypasses.[76] App-store billing systems are not merely payment rails. They integrate fraud detection, subscription management, and dispute resolution at scale.
Apple reports that it blocked more than $9 billion in fraudulent transactions over five years, including $2 billion in 2024, terminated 146,000 developer accounts, and identified 4.7 million stolen credit cards. Google Play Protect performs a comparable role, while Android-team analysis indicates that apps obtained outside curated stores are more than 50 times as likely to contain malware.[77]
Multisided platforms invest heavily in screening participants and excluding bad actors because user trust is part of the product.[78] Steering moves transactions out of that environment and into developer-specific systems of widely varying quality.
The available evidence remains immature. Steering regimes have operated only since 2024 in the United States, 2024–25 in the EU, December 2025 in Japan, and June 2026 in the United Kingdom. The relevant harms—including phishing during redirection, post-transaction data exfiltration, fraudulent refund disputes, and subscription traps—are diffuse, under-reported, and often slow to emerge.[79]
The Dutch experience illustrates the mechanism. After alternative payments were mandated for dating apps, reported dating-app fraud rose markedly, even if the evidence does not establish causation conclusively.[80]
The absence of large-scale, attributable failures within such a short period does not establish that the risks are negligible. The CR should not be drafted as though it does.
C. Interstitial Screens Should Inform, Not Merely Redirect
We support the CMA’s decision to permit a single, neutrally worded interstitial screen. The draft, however, treats that screen as tolerated friction rather than consumer protection.
Paragraph 7(a) limits the screen to information ‘strictly necessary to inform an End-user that they are moving outside of’ the in-app purchase system. Paragraph 8(c) separately prohibits ‘warnings, disclosures, or other messaging’ whose ‘purpose or effect’ is to discourage a steered transaction.
Read together, those provisions may bar truthful, material disclosures that platform-mediated refunds, centralised subscription management, and family controls do not apply off-platform. Accurate information may discourage some transactions, but that does not make it improper. A regime whose statutory objectives include trust and transparency should not condemn a disclosure merely because it affects conversion.[81]
Users who transact off-platform assume counterparty risk with developers they may not know. A brief, factual explanation of what changes is the minimum that informed choice requires.[82]
The drop-off evidence cited against interstitial screens warrants scepticism in both directions. Developers estimate conversion losses from a single neutral screen at anywhere from 5 per cent to more than 60 per cent. That range itself suggests conjecture. The CMA rightly acknowledges that developers ‘are unlikely to have tested the exact scenario’, so the figures reflect expectations rather than measurement.[83]
The broader empirical record points in the opposite direction. Mandated choice screens and disclosures have rarely changed user behaviour materially, as the EU’s browser and search-engine choice screens demonstrated over several years.[84] If interstitial screens barely affect behaviour, the case for stripping them of useful information to preserve conversion is weak.
The CMA should reject the further restrictions on which it seeks views, including time limits, frequency limits, and default-off settings.[85] Each would reduce the screen’s value as a consumer-protection tool without a demonstrated benefit.
The comparison with physical-goods apps also misses the point.[86] Those transactions occur outside platform billing by design and generally benefit from established card-network protections. Digital-content transactions present different risks, including minors’ purchases, subscription abuse, and refund disputes. That asymmetry supports disclosure rather than undermines it.
V. The CR Sacrifices Dynamic Competition for Static Gains
A final concern is dynamic competition. Returns above incremental cost are not a regulatory anomaly to be engineered away. They finance fixed-cost creation, in platforms as in intellectual property more broadly. Commissions on paid transactions support free distribution, developer tools, and the reduced small-business rates from which most UK developers benefit.
Compressing those commissions towards incremental cost would not merely transfer rents to large developers. It would invite rebalancing through device prices, developer terms, and investment in shared services. Much of that burden would fall on the free-app majority and on consumers. As ICLE argued to the U.S. Supreme Court, this is a zero-sum reallocation: gains to a small group of developers reliant on paid content come at the expense of the much larger group that is not.[87]
The platforms’ continued profitability does not answer the point. Investment decisions are made at the margin. A rule that caps returns on marginal steered transactions near incremental cost tells investors that much of the upside from improving the platform will accrue elsewhere.
The deeper problem is that cost-plus regulation optimises for the wrong kind of competition. Anchoring prices to incremental cost pursues static allocative efficiency by reducing margins on today’s transactions. But consumer welfare in mobile ecosystems has come largely from dynamic competition: new devices, new capabilities, new form factors, and the sustained reinvestment needed to produce them.
Consumers may not remember the year a commission fell by a few percentage points, but they remember the features that reinvestment made possible. Treating a rapidly evolving platform like a mature utility with predictable demand and little left to invent trades a large, uncertain dynamic gain for a small, contestable static one.
The consultation’s headline benefit—lower prices on steered transactions—reflects precisely that static lens. The CMA’s admitted difficulty in identifying a ‘reasonable rate of return’ exposes the same problem: a utility model is poorly suited to pricing innovation rents.
The proposal would also make the United Kingdom a global outlier. U.S. courts, including the Epic panel whose cost language the CR echoes, accept that Apple may charge a commission on linked-out purchases. Japan permits steered-transaction fees of 10 and 15 per cent. Even the DMA, at least initially, prescribed neither an LRIC methodology nor a market-power haircut.
The DMA’s record nonetheless illustrates the costs of regulatory maximalism: compliance costs far beyond the European Commission’s projections; features delayed or withheld from EU users, including Siri AI, iPhone Mirroring, and Google’s AI Overviews; and intended beneficiaries leaving the market or reporting 40–50 per cent declines in organic traffic following compliance changes.[88]
The CMA’s prioritisation principles require it to consider the Government’s strategic steer and the broader effects of its actions.[89] A first CR that announces UK-specific price regulation of platform access is not the signal a growth-focused competition regime should send.
VI. Recommendations and Conclusion
The CMA seeks views on the benefits, costs, and design of the proposed steering CR.[90] ICLE offers seven recommendations.
First, sever the CR. Proceed with the steering-rights core in paragraphs 1 to 8 and 14, subject to the amendments below, and withdraw paragraphs 9 to 13 as drafted.
Second, replace the proposed fee methodology with a ‘fair and reasonable’ obligation assessed ex post. That obligation should rest on three elements: a benchmarking safe harbour under which fees within the range charged by comparable platforms and other steering regimes are presumptively compliant; transparency requirements, including published fee schedules and advance notice of changes; and a defined escalation trigger under which the CMA would revisit fee regulation only if evidence gathered over a specified observation period showed that fees had made steering commercially unviable under metrics established in advance.
Third, if the CMA retains a fee methodology, it should amend it substantially. The value assessment should be primary, with cost serving as a cross-check rather than a ceiling. Paragraph 13(d)’s market-power adjustment should be deleted or defined operationally before it takes effect; a firm cannot comply with an obligation it cannot understand. The paragraph 12 cost base should reflect the integrated Mobile Platform the CMA designated. The phrase ‘to the satisfaction of the CMA’ should be replaced with an objective evidential standard. The attribution window should be lengthened, and the automatic-renewal carve-out should be removed or made presumptive rather than absolute.
Fourth, recalibrate the safeguards. Paragraph 3(b) should replace ‘strictly necessary’ with ‘reasonable and proportionate’. Its list of legitimate purposes should be non-exhaustive and expressly include privacy and consumer protection.
Fifth, retain a single neutral interstitial screen. Permit accurate, non-alarmist disclosures about protections that do not apply off-platform, and reject the proposed time limits, frequency limits, and default-off settings.
Sixth, resolve the remaining scope questions by confirming that steering may direct users to developer-owned destinations and serve transaction-completion purposes, while preserving the WebView prohibition and child-protection measures. The CR should also use outcome-based rather than pixel-level parity requirements.
Seventh, extend the implementation period beyond three months. The DMA’s compressed timelines provide a cautionary example.[91] Any fee obligations should take effect only after the steering-rights obligations, and the final decision should include a review clause requiring reassessment after a defined period against published consumer-outcome metrics, with express authority to narrow or revoke the CR.
This will be among the first substantive CRs imposed under the DMCC, and its design will help define the regime. Properly calibrated, the steering-rights core fits the participative and proportionate model the CMA has promised. Much of it is already emerging through legal and commercial developments worldwide, including in the United Kingdom.
Paragraphs 9 to 13 do not fit that model. They would commit the CMA, at the regime’s outset, to administering the price of platform access under principles that cannot yield stable answers, on an evidential record that assumes key conclusions, and under a review standard too limited to correct the errors likely to follow.
The DMCC promised greater contestability without the DMA’s collateral costs. The CMA should keep that promise.
[1] Competition & Mkts. Auth., Proposed Steering Conduct Requirement for Apple: Consultation Document (July 2026) [hereinafter Apple Consultation]; Competition & Mkts. Auth., Proposed Steering Conduct Requirement for Google: Consultation Document (July 2026) [hereinafter Google Consultation]. Responses to both consultations are due by 5 p.m. on 28 July 2026. Google Consultation ¶ 6.11.
[2] The draft conduct requirements and supporting analyses are materially identical. Unless otherwise indicated, these comments cite the Apple Consultation and provide parallel citations to the Google Consultation where the paragraph numbering differs. Consistent with the Competition and Markets Authority’s instructions, these comments respond to both consultations. See Google Consultation ¶ 6.11.
[3] See Geoffrey A. Manne, Dirk Auer & Mario A. Zúñiga, ICLE Comments to UK CMA on Competition in Mobile Ecosystems, Int’l Ctr. for L. & Econ. (12 Feb. 2025), https://laweconcenter.org/resources/icle-comments-to-uk-cma-on-competition-in-mobile-ecosystems [hereinafter ICLE Mobile Ecosystems Comments]; Geoffrey A. Manne, Dirk Auer & Mario A. Zúñiga, Comments of the International Center for Law & Economics on CMA’s Proposal to Designate Apple and Google with Strategic Market Status, Int’l Ctr. for L. & Econ. (20 Aug. 2025), https://laweconcenter.org/wp-content/uploads/2025/08/ICLE-CMA-Apple-Google-Designation-comments.pdf [hereinafter ICLE Designation Comments].
[4] Geoffrey A. Manne et al., Comments of the International Center for Law & Economics: Recent Developments in Relation to Apple’s and Google’s App Store Rules, Int’l Ctr. for L. & Econ. (22 Apr. 2026) [hereinafter ICLE App Store Comments], responding to Competition & Mkts. Auth., Views Sought: Recent Developments in Relation to Apple’s and Google’s App Store Rules (Mar. 2026) [hereinafter Call for Evidence].
[5] Apple Consultation, ch. 3, Proposed Steering CR ¶¶ 9–13; Google Consultation, ch. 3, Proposed Steering CR ¶¶ 9–13.
[6] Google Consultation ¶ 6.4 (seeking views on ‘the measures that Google rolled out in the UK on 30 June 2026’).
[7] Competition & Mkts. Auth., Digital Markets Competition Regime Guidance (CMA194, Dec. 2024) [hereinafter CMA194].
[8] Apple Consultation, Proposed Steering CR ¶ 10(a)(i).
[9] Id. ¶ 4.107.
[10] Id. Proposed Steering CR ¶ 12(b).
[11] Id. Proposed Steering CR ¶¶ 10(b)(i), 13(d)–(e).
[12] Id. Proposed Steering CR ¶¶ 9(a), 12(d); id. ¶¶ 4.150–4.151.
[13] Id. n.128.
[14] Apple Consultation ¶ 5.17; Google Consultation ¶ 5.15.
[15] Apple Consultation ¶ 4.2; Google Consultation ¶ 4.2.
[16] Apple Consultation, Proposed Steering CR ¶¶ 4, 14.
[17] Apple Consultation ¶¶ 2.6(c)(iii), 2.12(d).
[18] See ICLE App Store Comments, supra note 4, at 6; Google Consultation ¶ 6.4.
[19] Apple Consultation ¶ 4.151; Google Consultation ¶ 4.150.
[20] Lazar Radic, Steering in the Fog: The DMA and the Turn from Market Oversight to Market Ordering (ICLE White Paper No. 2026-06-15, 2026) [hereinafter Radic, Market Ordering].
[21] Geoffrey A. Manne et al., Response of the International Center for Law & Economics: Consultation on the First Review of the Digital Markets Act, Int’l Ctr. for L. & Econ. (24 Sept. 2025) [hereinafter ICLE DMA Review Response]; Radic, Market Ordering, supra note 20, at 31–40.
[22] Press Release, Eur. Comm’n, Commission Finds Apple’s App Store Rules Breach Digital Markets Act (23 Apr. 2025), https://ec.europa.eu/commission/presscorner/detail/en/ip_25_1085; see ICLE App Store Comments, supra note 4, at 5 (describing successive scrutiny of the Core Technology Fee, initial acquisition fee, and tiered Store Services fee).
[23] Digital Markets, Competition and Consumers Act 2024, c. 13, § 103 (UK) (providing that courts review appeals from Competition and Markets Authority decisions under the digital-markets regime using judicial-review principles).
[24] Comments of the International Center for Law & Economics on the CMA’s Draft Digital Markets Competition Regime Guidance 25, Int’l Ctr. for L. & Econ. (2024) [hereinafter ICLE DMCC Guidance Comments] (urging the Competition and Markets Authority to begin with narrower, individual conduct requirements so it could observe the relationship between each intervention and its effects).
[25] Apple Consultation ¶ 6.8(a); Google Consultation ¶ 6.9(a).
[26] Carl Shapiro & Hal R. Varian, Information Rules: A Strategic Guide to the Network Economy 3 (Harv. Bus. Sch. Press 1999) (‘Information is costly to produce but cheap to reproduce … cost-based pricing does not work: a 10 or 20 per cent markup on unit cost makes no sense when unit cost is zero.’).
[27] ICLE App Store Comments, supra note 4, at 4.
[28] Competition & Mkts. Auth., Strategic Market Status Investigation into Apple’s Mobile Platform: Final Decision (22 Oct. 2025); Competition & Mkts. Auth., Strategic Market Status Investigation into Google’s Mobile Platform: Final Decision (22 Oct. 2025); see Apple Consultation ¶¶ 4.7–4.8 & n.40.
[29] On two-sided platform pricing and how platforms divide surplus with complementors, see Jean-Charles Rochet & Jean Tirole, Two-Sided Markets: A Progress Report, 37 Rand J. Econ. 645 (2006); David S. Evans & Richard Schmalensee, Matchmakers: The New Economics of Multisided Platforms (2016).
[30] Epic Games, Inc. v. Apple, Inc., No. 23-16234, slip op. at 41 (9th Cir. 11 Dec. 2025).
[31] Brief of Former Antitrust Enforcement Officials as Amici Curiae in Support of Apple Inc.’s Petition for Panel Rehearing and/or Rehearing En Banc, Epic Games, Inc. v. Apple, Inc., No. 25-2935 (9th Cir. 12 Mar. 2026); Brief of the International Center for Law & Economics as Amicus Curiae in Support of Appellant, Epic Games, Inc. v. Apple, Inc., No. 25-2935 (9th Cir. 30 June 2025).
[32] Verizon Commc’ns Inc. v. Law Offices of Curtis v. Trinko, LLP, 540 U.S. 398, 407–08 (2004); Pac. Bell Tel. Co. v. linkLine Commc’ns, Inc., 555 U.S. 438, 452–53 (2009); see also United States v. Grinnell Corp., 384 U.S. 563, 570–71 (1966).
[33] Google Consultation ¶ 4.126.
[34] Google Consultation ¶¶ 4.125–4.132.
[35] Dr Rachael Kent v Apple Inc [2025] CAT 67, ¶ 631(11) (appeal pending); see Google Consultation ¶ 4.129.
[36] ICLE App Store Comments, supra note 4, at 4–5; see Daniel G. Swanson & William J. Baumol, Reasonable and Nondiscriminatory (RAND) Royalties, Standards Selection, and Control of Market Power, 73 Antitrust L.J. 1 (2005); Joseph Farrell, John Hayes, Carl Shapiro & Theresa Sullivan, Standard Setting, Patents, and Hold-Up, 74 Antitrust L.J. 603 (2007).
[37] Microsoft Corp. v. Motorola, Inc., 795 F.3d 1024 (9th Cir. 2015); Unwired Planet Int’l Ltd v Huawei Techs. Co [2020] UKSC 37; Optis Cellular Tech. LLC v Apple Retail UK Ltd [2025] EWCA Civ 552.
[38] Ohio v. Am. Express Co., 138 S. Ct. 2274, 2285–87 (2018); see also Geoffrey A. Manne, In Defence of the Supreme Court’s ‘Single Market’ Definition in Ohio v American Express, 7 J. Antitrust Enf’t 104 (2019); Andrei Hagiu, Proprietary vs. Open Two-Sided Platforms and Social Efficiency (AEI-Brookings Joint Ctr. for Regul. Stud., Working Paper No. 06-12, 2006).
[39] Apple Consultation ¶¶ 4.138–4.140.
[40] Id. ¶ 4.135.
[41] Apple Consultation, Proposed Steering CR ¶ 9(b); see ICLE App Store Comments, supra note 4, at 9.
[42] ICLE App Store Comments, supra note 4, at 6 & n.13.
[43] Apple Consultation ¶ 4.85(b); Google Consultation ¶ 4.86(b).
[44] Ohio v. Am. Express Co., 138 S. Ct. at 2289–90 (recognising that anti-steering provisions may serve procompetitive purposes by preventing free riding); Brief of the International Center for Law & Economics as Amicus Curiae in Support of Petitioner at 9–14, Apple Inc. v. Epic Games, Inc., No. 23-344 (U.S. 27 Oct. 2023) [hereinafter ICLE Supreme Court Brief].
[45] Digital Markets, Competition and Consumers Act 2024, c. 13, § 19(5) (UK) (‘The CMA may only impose a conduct requirement or a combination of conduct requirements on a designated undertaking if it considers that it would be proportionate to do so for the purpose of one or more of the following objectives: (a) the fair dealing objective, (b) the open choices objective, and (c) the trust and transparency objective, having regard to what the conduct requirement or combination of conduct requirements is intended to achieve.’).
[46] Apple Consultation ¶ 5.17; see Google Consultation ¶ 5.15 (same for Google).
[47] Apple Consultation ¶¶ 4.126–4.128; Google Consultation ¶¶ 4.124–4.128; see infra Section III.B.
[48] Apple Consultation ¶ 5.44; Google Consultation ¶ 5.41.
[49] Digital Markets, Competition and Consumers Act 2024, c. 13, § 19(10) (UK); Apple Consultation ¶ 2.11.
[50] Apple Consultation ¶ 5.42(a)(i).
[51] See Rochet & Tirole, supra note 29; Evans & Schmalensee, supra note 29.
[52] Ernst & Young & Copenhagen Econ., Study on the Application of the Interchange Fee Regulation (2020); Todd J. Zywicki, Geoffrey A. Manne & Julian Morris, Price Controls on Payment Card Interchange Fees: The U.S. Experience (ICLE White Paper, 2014); Julian Morris, Todd J. Zywicki & Geoffrey A. Manne, The Effects of Price Controls on Payment-Card Interchange Fees: A Review and Update (ICLE White Paper No. 2022-03-04, 2022).
[53] Zhuoxin Li & Gang Wang, Regulating Powerful Platforms: Evidence from Commission Fee Caps, 36 Info. Sys. Res. 126 (2025).
[54] Call for Evidence, supra note 4, ¶ 21; see ICLE App Store Comments, supra note 4, at 7–8.
[55] Apple Consultation, case study following ¶ 1.13; Google Consultation, case study following ¶ 1.15.
[56] Apple Consultation ¶¶ 2.12(a), 5.36–5.40; Google Consultation ¶¶ 2.12(a), 5.33–5.37.
[57] Apple Consultation ¶ 5.43; Google Consultation ¶ 5.40.
[58] Apple Consultation ¶ 5.29.
[59] Apple Consultation ¶ 5.16.
[60] Apple Consultation ¶¶ 4.93, 4.109, 4.116.
[61] Apple Consultation, Proposed Steering CR ¶ 12(d); id. ¶¶ 4.136, 4.150–4.153.
[62] Apple Consultation ¶¶ 5.8–5.9; Google Consultation ¶¶ 5.8–5.10.
[63] Apple Consultation ¶ 5.9.
[64] Apple Consultation ¶ 4.78; Google Consultation ¶ 4.79.
[65] Apple Consultation ¶¶ 1.17–1.19.
[66] See Frank H. Easterbrook, The Limits of Antitrust, 63 Tex. L. Rev. 1 (1984); Geoffrey A. Manne & Joshua D. Wright, Innovation and the Limits of Antitrust, 6 J. Competition L. & Econ. 153 (2010).
[67] Digital Markets, Competition and Consumers Act 2024, c. 13, §§ 22, 25, 84 (UK).
[68] Apple Consultation ¶ 4.153.
[69] Radic, Market Ordering, supra note 20, at 34–35 (discussing Eur. Comm’n Staff Working Document SWD(2026) 123 final (28 Apr. 2026)).
[70] Apple Consultation ¶ 3(b).
[71] Apple Consultation ¶¶ 4.41–4.42.
[72] Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 Sept. 2022 on Contestable and Fair Markets in the Digital Sector (Digital Markets Act), arts. 6(4), 6(7), 2022 O.J. (L 265) 1.
[73] Epic Games, Inc. v. Apple, Inc., 67 F.4th 946, 971, 985–86 (9th Cir. 2023), cert. denied, 144 S. Ct. 681 (2024).
[74] Apple Consultation ¶ 2.5(d).
[75] Apple Consultation ¶ 4.34.
[76] Apple Consultation ¶ 5.19; Google Consultation ¶ 4.19.
[77] Apple, App Store Prevented More than $9 Billion in Fraudulent Transactions (27 May 2025), https://www.apple.com/newsroom/2025/05/the-app-store-prevented-more-than-9-billion-usd-in-fraudulent-transactions; Suzanne Frey, A New Layer of Security for Certified Android Devices, Android Devs. Blog (25 Aug. 2025), https://android-developers.googleblog.com/2025/08/elevating-android-security.html.
[78] Evans & Schmalensee, supra note 29, at 138–39 (noting that quality control ‘requires platforms to invest significant efforts into investigating participants and kicking out bad ones’).
[79] ICLE App Store Comments, supra note 4, at 10.
[80] See ICLE App Store Comments, supra note 4, at 11 & nn.33–34 (discussing the Dutch Authority for Consumers and Markets’ alternative-payment mandate for dating apps and the subsequent rise in reported fraud).
[81] Digital Markets, Competition and Consumers Act 2024, c. 13, § 19(8) (UK); Apple Consultation, app. A, ¶ A.2(c).
[82] ICLE App Store Comments, supra note 4, at 8–9.
[83] Apple Consultation ¶¶ 4.61–4.62.
[84] ICLE Mobile Ecosystems Comments, supra note 3, at 10–12; Geoffrey A. Manne, A Critical Analysis of the Google Search Antitrust Decision 16–17, Int’l Ctr. for L. & Econ. (14 Aug. 2024).
[85] Apple Consultation ¶ 4.67.
[86] Apple Consultation ¶ 4.64.
[87] ICLE Supreme Court Brief, supra note 44, at 21.
[88] ICLE DMA Review Response, supra note 21, at 18, 23–25; Radic, Market Ordering, supra note 20, at 27, 34–35.
[89] Apple Consultation, app. A, ¶ A.5; Competition & Mkts. Auth., Prioritisation Principles (CMA188, Oct. 2023).
[90] Apple Consultation ¶ 6.2; Google Consultation ¶ 6.2.
[91] Apple Consultation ¶ 4.149; see ICLE DMA Review Response, supra note 21, at 18.