📲 A Fee That Finally Asked What I Made August 24, 2026 · https://tavi-blog.github.io/a-fee-that-finally-asked-what-i-made/

Apple announced last week that it's retiring the fee it has charged European developers since 2024 for distributing outside the App Store, and the mechanics of what's replacing it are more interesting than the headline number. The old fee was fifty cents per install, charged once a title passed a million installs in the EU in a year, and it didn't care whether a single one of those installs produced any revenue at all. A free app that got shared around enough could owe real money attached to nothing. Starting in October, that's gone, replaced by a flat five percent commission on transactions that actually happen. Money changing hands is now the trigger. Installs alone aren't.

I want to sit with why anyone built it the other way first, because it wasn't an accident or an oversight. Apple's argument for charging per install was that the App Store did real work before a single dollar showed up: it hosted the binary, ran it through review, made it discoverable, kept it updated and secure on someone's phone. That value gets delivered the moment someone installs the thing, whether or not they ever pay for it. Charging only on revenue, from that angle, lets a free app soak up distribution infrastructure indefinitely and pay nothing back for it, no matter how much of that infrastructure it actually used. It's not a crazy position. A lot of real infrastructure costs money to run regardless of whether the thing running on it makes anyone rich.

But it only holds up if you assume every install represents something the developer chose, or benefited from, in a way proportional to the number. I built a small tracking app a couple of years ago that never had a line of monetization code in it, not a subscription, not an ad, not an in-app purchase. Every install of that app produced exactly the same amount of revenue: none. Under a fee structured the old way, it wouldn't have mattered whether it stayed at a few dozen downloads or got picked up by the wrong algorithm somewhere and did the kind of number a free utility occasionally does by accident. The bill would have scaled with attention, not with anything I earned from that attention, and past a certain point I'd have owed money for the app existing and being liked, which is a strange thing to be billed for when nothing about how I built it or ran it had changed.

I never shipped into the market this specific fee governed, so it was never a bill I was at risk of actually getting. What stuck with me reading about the change wasn't relief on my own behalf. It was recognizing the shape of the risk from the inside, because I've watched install numbers move on a dashboard for an app that owed nobody anything, and I know exactly how little control a solo developer has over whether that number goes viral for a week. You don't get to decide when a screenshot gets reshared somewhere you didn't post it. A fee formula that treats that kind of luck as taxable value, indistinguishable from a business actually converting attention into money, is a formula that punishes the exact kind of small, free, unmonetized project a person builds at night for the experience of shipping something.

The percentage model fixes that specific failure mode without pretending the underlying argument about infrastructure cost was wrong. A five percent commission on transactions still lets Apple bill for the App Store's work, just tied to the moment that work actually translated into money for someone. An app that stays free forever owes nothing forever, which is the correct answer for that app. A subscription business that scales pays proportionally more as it scales, which is also arguably the correct answer, and is a heavier ask of the businesses actually generating the revenue in the first place rather than a flat tax that landed hardest on whoever had the least ability to absorb a surprise.

What I keep coming back to is how long the install-based version stood before regulatory pressure, not competitive pressure, forced the redesign. Nobody inside Apple was going to notice on their own that fifty cents times an unpredictable number is a strange way to price something for the person on the other end of it. The formula worked fine from the side that was collecting. It only looked broken from the side holding a dashboard, watching a count go up, and doing the arithmetic on what happens if it doesn't stop.