If you sell digital goods in an app to UK users, the short version is this: the UK's Competition and Markets Authority closed its consultation on steering at 11:59pm on 28 July 2026, a decision is expected later in 2026, and there is nothing to wait for before doing the work. Google has already published the UK fee schedule that applies from 30 June 2026, and it contains a distinct, lower rate for purchases completed through an external web link. That rate is the number to model against, and the link-out page is the thing to build while the regulator writes its decision.
What the CMA proposed, precisely
Two parallel consultations ran from 8am on 30 June 2026 to 11:59pm on 28 July 2026, one for Apple's mobile platform and one for Google's. Both follow the strategic market status designations made in October 2025. The Google consultation page now reads "We are analysing your feedback."
The proposed conduct requirement would let developers "engage with customers about off-platform options," which the CMA's announcement says is currently banned by Apple and restricted by Google in the UK. Three details in that announcement decide how much this is worth to you:
- Steering is about being allowed to tell users about, and send them to, a cheaper route, not about a free ride. Apple and Google may still charge.
- Any steering fee must be "justified through a robust, evidence-led framework involving due reference to both cost and value," and the CMA expects such fees to be substantially lower than current app store charges.
- A separate proposed requirement would give developers access to Apple's NFC functionality on iOS; responses on that closed earlier, on 21 July 2026.
What the CMA has not published is a number. There is no proposed percentage in the announcement, and the outcome documents are not out. Anyone quoting you a specific UK steering fee today is guessing.
Google already repriced, and the external-link rate is public
This is the part most coverage skips. Google Play's service fee page sets out a separate schedule for the EEA, the UK and the US, effective 30 June 2026, that splits fees by install cohort and by whether the transaction runs through Play billing or an external web link.
| UK, EEA and US, from 30 June 2026 | First $1M a year | Above $1M |
|---|---|---|
| New installs, Play billing | 10% + 5% billing fee | 25% + 5% billing fee |
| New installs, external web link | 10% | 20% |
| Existing installs, Play billing | 20% + 5% billing fee | 25% + 5% billing fee |
| With Play Games Level Up or Apps Experience | 15% + 5% billing fee, or 10% for external web links | |
| All other markets (unchanged) | 15% first $1M, 30% above; auto-renewing subscriptions flat 15% | |
Read the "+ 5% billing fee" as the separable part. That is what an external web link removes, and it is why the external column sits five points below the Play billing column at each tier. The 5% is not the whole prize, though, because you then pay your own processor, and card processing plus tax handling is rarely free.
The arithmetic, at two revenue levels
Our math, using the published rates above and a card cost of roughly 3% plus fixed fees for a self-run checkout. Assume a new-install UK user base and annual revenue below the $1M threshold.
| Annual UK revenue | Play billing (10% + 5%) | External web link (10% + ~3% card) | Difference |
|---|---|---|---|
| $60,000 | $9,000 | $7,800 | $1,200 |
| $400,000 | $60,000 | $52,000 | $8,000 |
Two caveats that change the answer for real businesses. First, the card cost is a placeholder: your actual rate depends on your processor, your mix of international cards and whether you use a merchant of record, which we compared in Paddle's 5% or Stripe's 3.4%. Second, a link-out checkout adds work you do not currently do: VAT handling, refunds, receipts, chargebacks and subscription dunning. At $60,000 a year, $1,200 does not obviously pay for that. At $400,000, $8,000 starts to.
The number to watch is the cohort split. The published schedule charges 20% + 5% on existing installs in the first $1M band against 10% + 5% for new ones. If most of your UK revenue comes from users who installed before the change, your effective rate is double the headline you may have read, and the case for building a link-out route is correspondingly stronger.
What to build before the decision lands
- Split your UK revenue by install cohort and by product type. One spreadsheet, four cells: new-install subscriptions, new-install one-off, existing-install subscriptions, existing-install one-off. Apply the rates in the table. You now know your blended UK take, which is the only number that makes the rest of this decision.
- Stand up a web checkout that works without the app. Not a link that opens a broken mobile page: a real product page, real prices in pounds, a checkout that completes on a phone, and an account system that recognises the same user inside the app. This is the long-lead item and it is entirely within your control today.
- Decide the entitlement path before the payment path. When someone buys on the web, the app has to unlock the feature. Server-side entitlements keyed to your own user ID, checked on launch, are the version that survives platform changes. Receipt validation against store APIs is the version that does not.
- Write the in-app copy now and keep it in a feature flag. When steering is permitted, the difference between a 2% and a 20% take-up is the sentence you show and where you show it. Draft it, get it reviewed, ship it disabled.
- Keep store billing as the default until the fee decision is published. Removing in-app purchase before you are allowed to steer is how apps get rejected. The levers available when review says no are set out in the rejection playbook, and none of them beats not triggering the rejection.
How this fits what has already happened elsewhere
The UK is the third regime to arrive at the same structure by a different route. The EU's approach produced a fine over steering restrictions rather than a fee schedule, and the commission tiers small developers actually pay are the subject of the 15% you have to ask for. The consistent pattern across all three: the platform keeps a fee even when it does no payment processing, and the argument is only ever about how large that fee is. Plan on paying something, and size the build accordingly.
Concretely, the calendar for the rest of 2026: the CMA is analysing responses now, its decision on whether to impose the requirements is expected later this year, and Google's UK fee schedule is already live. If you do the cohort split this week and have a working web checkout by the time the decision publishes, you will be able to act on the rule in days rather than quarters. If you wait for the number, you will spend the quarter after the announcement building what you could have built during the consultation.
Discussion
Sign in with Google or just a name. No email link, no password to remember.