Usage based billing on Google Play turns the subscription into a meter

Usage based billing on Google Play: what actually shipped

A fox learns a thicket by walking it, so start with the trail Google has actually laid. On 29 September 2026, 9to5Google reported that Google Play is giving Android developers more payment options around subscriptions, and the headline act is usage based billing. Google's own Android Developers Blog post describes it as prepaid metered billing: users top up a balance automatically whenever it falls below a set threshold, which Google says keeps service uninterrupted for users while protecting a developer's margins.

Three companions arrived in the same post. Multi-quantity subscription purchase lets one buyer make several subscription purchases in a single transaction and assign them as seats to team members or students. Mixed carts let a developer sell an auto-renewing subscription and one-time products in one checkout. Cross-developer bundling lets a developer create a hard bundle of two or more complementary subscriptions, including a partner's, as a single product at a discounted rate.

The pitch is aimed squarely at generative AI. Google says the subscription landscape is shifting with the surge of generative AI, and Android Authority summarised the change as usage based subscription billing for AI apps and other services with variable costs. If your quarry is an app whose running cost rises with every prompt, image or minute of compute, this is a Play-native billing shape built for your problem.

Be careful about what is not yet on the record. Google's post is a product announcement, not a rate card. In what we read, it publishes no pricing for usage based billing, no regional availability list and no reporting detail, and heise online describes Google as currently testing the tools. Treat everything below as a planning framework for something in early access, not a settled spec.

Metered billing and your LTV maths: the flat line becomes a wobble

Start with the maths you already do. For a flat plan, lifetime value is a tidy little sum: price per period, times the number of periods a subscriber stays, less the store's cut. One number goes in, one curve comes out, and your cohort charts are smooth enough to frame. That tidiness is the thing a metered billing model takes away.

Under usage based billing, a customer's value is the sum of every top-up across their life: how big the top-ups are, how often the balance runs low, and how long the customer stays. Each of those is a distribution, not a constant. A few heavy users will fund the whole cohort while a long tail tops up once and drifts off. Averages stop describing anyone, which is why a vulpine growth team prowls the distribution instead.

• 5%

  Play billing fee in the US, UK, EEA, Australia and Japan

• 41%

  more revenue per payer for AI apps, per RevenueCat

• 30%

  faster churn for AI apps, per RevenueCat

• 55%

  of 3-day trial cancellations land on day zero

The store's cut also has a cohort shape. Google's help page on lower service fees says the applicable service fee is determined by whether the transacting user's install is new or existing, and that purchases through Google Play Billing carry an additional 5% billing fee in Australia, Japan, the US, the UK and the European Economic Area. The service fees page gives the rollout dates as 30 June 2026 for the EEA, UK and US and 30 September 2026 for Australia and Japan. So net LTV should be modelled by install date, not by one blended percentage.

Now the cost side, which flat plans let you ignore. RevenueCat's State of Subscription Apps 2026, built on more than 115,000 apps and over $16 billion in revenue, reports that AI apps earn 41% more revenue per payer but churn 30% faster. Phil Carter of Elemental Growth, quoted in the report, adds that the cost to serve a marginal subscriber used to be near zero, and that this is not true for products running large language models. That sample is RevenueCat's own customer base, so read it as a strong signal, not a census.

There is a behavioural twist too. In a 2006 study of internet access tariffs, Lambrecht and Skiera found that many customers prefer a flat rate even when pay-per-use would cost them less, and that the flat-rate bias does not significantly increase churn. They also found that the pay-per-use bias, where people choose metering and then underestimate their usage, largely increases churn. That is a decades-old finding from a different product, so it is a hypothesis for your app, not a Play result, but it is exactly the hypothesis worth testing.

Hold the usage based pricing model against your own users before you hold it against a spreadsheet. Ask which segment will under-estimate its usage, because that is the segment most likely to feel surprised by its third top-up and leave. The ones who top up, shrug and carry on are your quiet profit. Cohorts first, averages last.

Paywalls, trials and ASO when price is not one number

Those gaps are the old levers, and metered billing adds a third: how big a first top-up you ask for. Every decision on that chart has a metered twin. A free trial becomes a free starter balance. A hard paywall becomes a minimum first top-up. A plan comparison becomes a conversation about what a unit buys. RevenueCat reports that trials of 17 days or more convert at a median 42.5% against 25.5% for short ones, and that 55% of three-day trial cancellations happen on day zero, so the first session still decides who stays.

Framing matters more than price here. In the same report, Michal Parizek of Mojo describes a test in parts of Latin America where anchoring the yearly plan to its monthly equivalent lifted trial starts by 30% with no impact on trial-to-paid conversion, a single-app result, not a benchmark. When a user cannot see one price, the unit of account is your messaging. Say what a top-up buys in plain terms (minutes, images, pages), not in abstract credits, and show the balance before you ask for the next payment.

Google's rules put guard rails on that copy. Its Payments policy says developers must clearly and accurately inform users about the terms and pricing of in-app features and subscriptions, that in-app pricing must match the Play billing interface, and that a listing must say when features need an extra charge. A paywall that hides the auto top-up threshold is the sort of thing that policy exists to stop, and a pricing page that disagrees with the checkout sheet is the sort of undergrowth that gets apps flagged.

ASO has a rule of its own. Google's metadata policy says not to use text or images in the app title, icon or developer name that indicate price or promotional information. So the metered pitch belongs in your description, screenshots and in-app flows, not in your title. Our recent notes on Google Play ASO and short descriptions and creative readiness at launch cover the craft of writing for those slots.

Trial design needs a rethink as well, because a metered user can churn without ever cancelling: the balance simply stops being topped up. Define churn for metered cohorts as a gap, such as no top-up and no usage for a set number of days, and keep it separate from your subscription cancellations. Otherwise your retention dashboard will report health while the quiet undergrowth fills with lapsed balances.

Your app subscription model: bundles, seats and mixed carts as growth levers

• Usage based billing

  Prepaid metered billing

  How it bills

  Auto top-up below a threshold

  Aimed at

  Variable computing costs

  Best for

  • AI generation tools
  • Pay as you go services

• Multi-quantity

  Seats in one purchase

  How it bills

  Several subscriptions, one transaction

  Aimed at

  Teams, classes, groups

  Best for

  • Productivity and EdTech
  • Team access for AI tools

• Mixed carts

  Subscription plus one-time items

  How it bills

  One checkout, one API call

  Aimed at

  Upsell at the point of sale

  Best for

  • Starter packs with a plan
  • Bundle discounts

• Cross-developer bundles

  Hard bundle of subscriptions

  How it bills

  One product, partner catalogue

  Aimed at

  New audiences via partners

  Best for

  • Complementary apps
  • Portfolio cross-sell

Run your eye along those cards as a growth lead and each one is an acquisition or retention lever dressed as a billing feature. Google's post pitches cross-developer bundling as a way to reach new audiences by teaming up with other developers or combining your own apps, with a language app and a travel guide as its example. Mixed carts, it says, open upsell routes such as targeted discounts when a user buys a subscription and complementary in-app items together.

Those are paid-media questions as much as product ones. A bundle partner can be a distribution channel with no CPI line, which makes attribution the sticky part: who gets credit when a user arrives through a partner's subscription? Our piece on mobile measurement and AI-answer attribution is a useful primer on trusting the method before you trust the number, and our app user acquisition strategy note covers where the budget goes first.

Seats change the buyer. When one purchase covers a team or a class, your acquisition target moves from the individual to the administrator, and your creative should speak to the person who assigns the seats. That is a different message, a different channel mix and a different measure of success, so build a separate funnel for it rather than stuffing it into the consumer one.

Which brings retention back to the centre. Google's post also lists tools that sit around the new billing shapes: an in-app messaging API for payment declines and price changes, a dynamic grace period that tailors the recovery window to each subscriber, retention offers shown in the Play cancellation flow, and winback offers that reach lapsed users directly in the Play Store. Google says several of these are in the same early access programme, so check what your account can switch on. Our 2026 app retention playbook and a recent look at retention and remarketing leaks are good companions.

Any app subscription model that adds a bundle, a seat count or a meter also adds fee questions. Our earlier read on the store fee squeeze and on how storefront price changes reach your margin explain why net revenue per user, not list price, is the number your acquisition bids should chase.

How to test usage based billing without betting the den

Nobody can outfox a pricing change by theorising about it. The cleanest approach is a patient, staged test that treats usage based billing as a hypothesis about your own users, with Google's early access limits as the fence around it. You are looking for four answers: do new users top up at all, how often do they come back, what does a cohort cost to serve, and who leaves quietly.

Keep acquisition and billing tests in separate windows so you can tell which one moved the number, and give the pilot a guardrail: refunds, support tickets and failed top-ups should not climb. If your acquisition runs on paid media, our app marketing team can help map the funnel, and if the product itself is an AI feature, the AI consultancy lane covers evaluation and unit cost.

A last scent worth following is regulatory. The UK government announced on 9 August 2026 that, per its press release, new subscription rules come into force in January 2027, bringing clearer up-front information, regular reminders and a much easier exit. Whether a prepaid balance that tops itself up counts as a subscription contract is a question for your lawyers, not for us, but design the consent and the cancel path as if it might.

So the plan is plain. Rebuild LTV as a distribution, model fees by install date, write paywall copy in units a user understands, define churn for metered cohorts, and pilot small. If you want a second pair of eyes on any of it, start a conversation and we will walk the hedgerow with you.

Frequently asked questions