App monetization strategy, the five models and the store fee arithmetic that decides which one survives

There are five ways a consumer app earns money. A one-off purchase, a subscription, in-app purchases of consumable or unlockable items, advertising, and selling the app as a route into a business that charges somewhere else. Picking between them is not a taste question, it is arithmetic with two inputs. Your retention curve, which decides whether anything recurring is possible at all, and the store's cut, which is a larger and more structured number than most plans assume. Google Play's own service fee page, read on 31 October 2026, now sets the fee by whether a paying user's install is new or existing, with regional rollout dates of 30 June 2026 for the EEA, the UK and the United States and 30 September 2026 for Australia and Japan. That means the same transaction in the same app can carry a different fee depending on when the user first installed it, which is not a detail any model can be built without.

The store cut, as published on 31 October 2026

Take these as a dated snapshot of two developer pages and re-read them on the day you model anything, because this is the most frequently changed number in the whole category.

Google Play, for transactions with users in Australia, the EEA, Japan, the United Kingdom or the United States. The fee depends on the fee tier and on whether the user's install is new or existing, where a new install means a first-time install or first update on or after the regional rollout date.

	Auto-renewing subscriptions, new and existing installs	Other transactions, new installs	Other transactions, existing installs
First 1 million USD of annual earnings	10 per cent plus a 5 per cent billing fee	10 per cent plus a 5 per cent billing fee	20 per cent plus a 5 per cent billing fee
Standard	10 per cent plus a 5 per cent billing fee	20 per cent plus a 5 per cent billing fee	25 per cent plus a 5 per cent billing fee, or 20 per cent for external web links
In the Play Games Level Up or Apps Experience programs	10 per cent plus a 5 per cent billing fee	15 per cent plus a 5 per cent billing fee	20 per cent plus a 5 per cent billing fee, or 15 per cent for external web links

Google's note on the billing fee is that it "applies when a user completes a purchase of an in-app digital service or feature using Google Play Billing or for the initial purchase of a paid app or game". For all remaining markets, until the updated fees roll out globally, Google publishes the earlier structure: developers in the 15 per cent tier pay 15 per cent on the first 1 million USD of annual revenue and 30 per cent above it, while automatically renewing subscriptions are 15 per cent regardless of revenue. Where an alternative billing system is offered to users in South Korea or India under the Payments policy, Google states the fee for those transactions is the applicable Play fee reduced by 4 per cent. Google also states that 97 per cent of developers distribute at no charge, and that of those subject to a fee, 99 per cent are eligible for 15 per cent or less.

Apple's App Store Small Business Program, read the same day, publishes a reduced commission rate of 15 per cent on paid apps and Apple In-App Purchases. Eligibility is stated as existing developers who made up to 1 million USD in proceeds in the prior calendar year across all their apps, and developers new to the App Store. If a participating developer passes the 1 million USD threshold in the current calendar year, Apple states the standard commission applies to future sales, and a developer whose proceeds fall below it in a future year can re-qualify the year after. Associated developer accounts are counted together. Apple also states that for developers on the alternative terms in the EU who are in the programme, and for subscriptions after their first year, it offers a further reduced commission of 10 per cent. Apple's standard commission rate outside that programme was not read on 31 October 2026 and is not quoted here, so take it from Apple's own page before modelling.

The practical reading of all of that is one sentence. Subscriptions carry the lowest published rate on both stores at the small-business end, and under Google's current table a new install paying for anything is cheaper to serve than an existing one in the markets that have rolled over. Both of those facts push in the same direction and neither was true two years ago.

The retention curve, which eliminates three of the five

Before choosing a model, plot day one, day seven and day thirty retention for your existing users. That curve decides what is available to you.

• If day thirty retention is low, a subscription is not available and neither is advertising. A subscription needs people still opening the app at renewal, and advertising needs enough sessions per user to accumulate impressions. What is left is a one-off purchase or a single early in-app purchase, which means all of your revenue has to be collected in the first week.
• If day thirty retention is healthy and sessions are short, advertising becomes viable, because impressions come from frequency rather than from session length.
• If day thirty retention is healthy and the app is used for a job rather than for entertainment, a subscription is the strongest option on the table, and it is the one with the lowest published store fee.

This is why monetisation decisions made before launch are usually wrong. The curve is not knowable in advance, so the better sequence is to ship, measure thirty day retention on real users, then choose.

The five models, and what each one really requires

Model	What it needs to work	Where it breaks
One-off purchase	A clear, demonstrable benefit before download, and no ongoing cost to serve	No recurring revenue, so every month starts from nothing and acquisition cost has one shot to pay back
Subscription	Value that recurs, and a reason to open the app in week four	A subscription on a one-off job, which produces excellent first month numbers and brutal churn
In-app purchases	A small share of users who will spend repeatedly, and a design where paying is a choice rather than a toll	Revenue concentrated in very few users, which is a fragile base and a design pressure in a direction users resent
Advertising	High session frequency and enough volume for the inventory to be worth buying	Small audiences, where the revenue per user is a rounding error against the harm to the experience
App as a route to a business	Something to sell elsewhere, such as a service, a physical product or a web subscription	Store policy, which governs what may be sold where. Read each store's rules before designing the flow

The fifth is the one most often overlooked by small teams and it is frequently the right answer. An app that generates no revenue inside the store and books appointments, sells goods or retains customers for a business that earns elsewhere has sidestepped the fee question entirely, and the arithmetic is the business's arithmetic rather than the store's.

Three ways the arithmetic gets done wrong

• Modelling on list price instead of net. A subscription at a given monthly price nets the price less the store fee, less local sales tax or VAT where that is handled by the store, less payment and refund losses. Model the net figure from the first spreadsheet, not as an adjustment later.
• Comparing lifetime value to acquisition cost without a payback period. A twelve month lifetime value against a one month acquisition cost is a viable business only if you can fund eleven months of working capital. Write the payback month down, because that is the number that decides whether growth is fundable.
• Averaging across both platforms. The fee structures differ, the user behaviour differs, and under Google's current table the fee can differ by install age within one platform. Model each store separately and expect the mix to change the blended rate as the user base ages.

Questions about app monetization

Which app monetization model earns the most?

Subscriptions, where the app genuinely delivers recurring value, because revenue compounds and the published store fee on auto-renewing subscriptions is the lowest tier on both stores. Where retention past thirty days is weak, a one-off purchase or an early in-app purchase is the realistic model.

What does the app store take?

Read off the two developer pages on 31 October 2026: Google Play sets a tiered service fee plus a 5 per cent billing fee in the rolled-over markets, from 10 per cent on auto-renewing subscriptions up to 25 per cent on some existing-install transactions, and Apple's Small Business Program publishes 15 per cent for qualifying developers with a further reduction to 10 per cent in specified EU and subscription cases. Both change, so take the current figures from the stores before modelling.

Should a new app choose advertising or a subscription?

Measure thirty day retention first. Advertising needs frequency and volume, a subscription needs a reason to return in week four, and neither is knowable from a plan. Ship, read the curve, then decide.

Is a free trial worth offering?

It raises the number of people who start and it also raises churn at the trial end, so it is only worth it where the product proves itself inside the trial. Judge it on paid conversions at day sixty rather than on trial starts.

Can I avoid store fees by charging on my own website?

Whether and how you may direct users outside the app is governed by each store's current rules and has changed repeatedly, including the external web link rates in Google's own published table. Read the current policy for both stores before designing the flow rather than copying a competitor.

How many revenue models should one app use?

One primary model, and at most one secondary. Two models competing for the same decision confuses the user and makes the reporting unreadable, which is how teams lose track of which one is actually paying for the business.

The part of growth that is not in the store dashboard

Every number above is about what happens after somebody installs. The step before it has quietly changed. People looking for an app increasingly ask an assistant rather than browsing a store category, in words like what is the best app for tracking inventory in a small warehouse, and get back two or three app names with a sentence each. That shortlist is not built from your store ranking or your review count. It is built from what has been written about your app anywhere that could be read, which for most small teams is a thin store listing and nothing else. The effect on the arithmetic on this page is direct, because acquisition cost is the other half of every model here, and a named recommendation costs nothing per install. It is worth running the check before the next growth budget is set: ask the question your user would ask in two or three assistants, with your own app's name left out of the prompt, and write down which apps come back. AI Knows Us takes those readings across ChatGPT, Gemini, Perplexity, Claude and Google's AI Overviews and reports which sources each answer was built from, which tells you what is missing rather than only that you are absent.