US link-outs after the Epic ruling: what changed for the apps we buildUS link-outs after the Epic ruling: what changed for the apps we build

In December an appeals court modified the injunction from the Epic case, restoring Apple's ability to charge a fee on purchases made through external payment links in US apps.

Coverage framed it as a reversal of developer gains. For the apps we actually build, the practical position barely moved - and the reason is worth setting out, because it applies to the next ruling too.

Most apps were never going to take the exit

The window in which external links carried no Apple fee looked like a large opportunity. Taking it required building and operating a payment business: checkout, receipts, refunds, chargebacks, fraud screening, sales tax registration across jurisdictions, and the support load attached to all of it.

For a team of five shipping a subscription app, that is not a saving. It is a second product with no revenue of its own.

The costs that do not appear in the commission comparison

The arithmetic people run is the platform commission against a payment processor's percentage. That comparison leaves out most of the work.

• Tax. In-app purchase handles registration, collection and remittance across every market you sell in. Doing it yourself means either a tax service with its own fee or an obligation you have taken on without noticing.
• Refunds and chargebacks. Someone has to decide, process and account for them, and users will ask you rather than the processor.
• Fraud. A card checkout attached to a consumer app attracts card testing within weeks. Screening is not optional and is not free.
• Recovery. Failed renewals are a meaningful share of subscription revenue. Retry logic, dunning email and card updates are an entire discipline that the platform currently performs silently.
• Support. Every one of the above generates messages from people who are confused, and confusion about money escalates faster than confusion about anything else.

None of this argues that external payment is wrong. It argues that the comparison is not a percentage against a percentage. It is a percentage against a percentage plus a team.

What we advised at the time, and still do

• If you already run web billing at scale, evaluate it seriously - the marginal cost is genuinely low.
• If you do not, the commission is buying infrastructure you would otherwise build badly.
• Either way, do not architect for exactly one payment route.

That last point is the durable one. The commercial terms here have changed three times in two years across two jurisdictions. Any design that assumes the current rules are permanent will be revisited.

Build for the rules changing, because the rules keep changing.

What replaceable actually means in code

Keeping the payment layer replaceable sounds like an architecture project. It is mostly a discipline about where a few facts are allowed to live.

• Entitlement is yours, not the processor's. The app should ask your own backend what this user is allowed to do, and that answer should not name a store or a card.
• Product identifiers belong to you. Mapping your plan to a platform product is a translation at the edge, not an identity used throughout the app.
• Receipts are an implementation detail behind one verification service. Nothing above it should know what a receipt looks like.
• Purchase user interface is separate from purchase logic. The paywall changes per jurisdiction and per ruling far more often than the rules about what a subscriber gets.

A team that has those four in place can add a second payment route in weeks and turn it off again just as fast. A team that has entitlement scattered through feature code cannot respond to any ruling, in either direction, without a rewrite.

The part that did change

What genuinely shifted is the negotiating position. There is now a documented path to selling outside in-app purchase, at a known cost, with the platform's acknowledgement. That matters for large publishers with real leverage.

For a fifteen-person company it is mostly a reason to keep an eye on the terms and keep the payment layer replaceable. Which was the right answer before the ruling as well.

When it is worth revisiting

Rather than re-running this analysis after every headline, we suggest clients pick the conditions that would change the answer and ignore the news until one of them is met.

• Subscription revenue reaches the point where a few percent pays for a person to operate billing.
• You already sell the same product on the web to the same users, and the second system is a route rather than a new build.
• A jurisdiction you actually sell in mandates a change, rather than one you read about.

Until one of those is true, the commission is the cheapest billing team available, and attention is better spent on the product. When one becomes true, the four disciplines above are what decide whether the switch takes a fortnight or a quarter.