Regulation lands on the buyer, not the platform

Every antitrust ruling and platform regulation of the past five years has named Apple or Google as the party under discipline. The studios buying installs paid for every one of them, in working hours as well as in deferred decisions. This isn't a complaint about fairness. It is a structural feature of how platform regulation works, and a UA team that doesn't budget for it will keep getting surprised by costs it can't trace to any line item.

A thoughtful head of platforms could disagree. Regulation has delivered real gains: link-outs in the US, alternative distribution in the EU, a fee structure on Play that would have been unthinkable in 2023. The trend is towards more room for developers, not less. All true, and none of it changes who does the implementing.

The regulated party controls the implementation

The mechanism is simple once stated. A court or regulator sets a requirement; the platform, as the regulated party, chooses how to comply. It complies in the way that best preserves its economics, which almost always means a mechanism that satisfies the letter of the requirement while pushing complexity to the edge of the system, where the developers are.

Apple's response to the Digital Markets Act is the cleanest recent illustration. At the start of January Apple replaced the Core Technology Fee, a per-install charge, with a five percent Core Technology Commission on digital goods for EU developers on standard terms, according to its own developer documentation. The Coalition for App Fairness, in its public response, maintained that the new terms still did not comply. Whether they do is for the Commission to decide. What is certain is that a studio operating in the EU has now modelled its unit economics under at least three fee regimes in eighteen months, and each remodelling was work the studio did, not Apple.

The same shape appears in the US link-out saga. The 2025 contempt ruling forced Apple to allow external payment links without commission. Apple appealed. In late April the Ninth Circuit reversed the stay and sent the fee question back to Judge Gonzalez Rogers, as the trade press reported at the time. Earlier this month Epic filed asking the Supreme Court to deny Apple's petition. Every one of those dates changed the expected value of building a web shop, and every studio weighing that investment has had to re-run the decision each time. The platform's legal team files a brief. The studio's growth team re-plans a roadmap.

Three questions to ask of every ruling

When a regulatory event lands, the coverage will tell you who won. That is the least useful thing to know. Ask instead who implements, who bears the uncertainty, and who captures the surplus.

Who implements? If the answer is "the platform, through a mechanism developers must adopt", the cost is yours. Nearly every major ruling since ATT fits this description.

Who bears the uncertainty? A ruling under appeal is not an outcome; it is an option with an unknown expiry. Whoever has to make investment decisions before the appeal resolves is carrying the risk, and that is the developer, not the platform, which has the balance sheet to wait.

Who captures the surplus? Where a fee is cut or a channel opened, the gain is real but typically ends up shared with whoever the developer must now pay instead: the payment processor, the web shop vendor, the alternative store. The platform's loss is rarely the developer's gain in full.

Run the Ninth Circuit remand, the WWDC announcements from Monday's keynote and the Supreme Court's pending cert decision through those three questions and the pattern holds each time. WWDC, for what it is worth, was about store bundles and discovery plus asset workflows; the early recaps from the attribution vendors report no change to AdAttributionKit or SKAN, which is the rare case where the answer to all three is "nobody, yet".

The second-order effect: regulation consolidates the market it meant to open

Compliance is a fixed cost. Modelling a new fee regime, building a link-out flow, submitting to an alternative store's review process, updating consent language: none of these scale with revenue. A publisher with fifty live titles spreads the cost across fifty; a studio with two titles carries it on two.

The consequence is uncomfortable for anyone who cheered the rulings. Regulation intended to reduce the platforms' power over developers has raised the minimum viable size of a developer who can exploit the new freedoms. The studios adopting web shops most aggressively, along with alternative distribution and link-out payments, are the ones that needed them least, because they are the only ones with the engineering and legal capacity to absorb the implementation. The long tail continues to pay full commission, not because the option is shut to them but because the option costs more to exercise than it returns at their scale.

There is a second effect on timing. Uncertainty itself is a cost, and it is a cost the platforms can afford to impose for years. A studio that defers a web-shop build pending the Supreme Court is behaving rationally and is also forgoing the margin the ruling was meant to deliver. The longer the appeal runs, the longer the platform collects at the old rate from everyone who waited. Delay is not a side effect of the platform's legal strategy. It is the strategy.

What this means for a UA budget

Two practical changes follow.

First, create a regulatory implementation line in the growth budget and fund it. An illustrative figure: a mid-sized studio might allocate the equivalent of one engineer-month per quarter plus a modest legal review retainer. Most quarters it will be underspent. In the quarters that need it, it stops the work coming out of the campaign roadmap, which is where it currently comes from without a budget or a record.

Second, treat every ruling as a dated option, not an outcome. Write down what the ruling permits, what it would take to exercise it, what the exercise would return per month, and the date by which the appeal or enforcement question resolves. Then compute the break-even: how many months of the new terms do you need to recover the implementation cost? If the appeal is likely to run longer than the break-even, the rational move is to prepare the build and hold it, not to ship it. Where the break-even is short you ship and accept the risk of a reversal.

An illustrative example: a studio estimates a US web shop costs three engineer-months and returns a commission saving of a modest five-figure sum per month at current revenue. Break-even is around eight months. The Supreme Court's cert decision is due by the end of this month; if it takes the case, merits will run into next year. A studio with that arithmetic builds now and switches on only for the highest-value cohorts, where the saving per user justifies the friction, and keeps the rest on the store until the option is no longer under appeal.

The line item that has always existed

None of this cost is new. Studios have paid it every quarter since 2021 in roadmap slippage and in decisions deferred for a court date. Giving it a name and a budget does not make it larger. It makes it visible, and visible costs are the only ones that get managed.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• How platform fee changes flow into bids
• The UA lead as translator, not buyer