Most mobile game UA advice assumes a player who watches ads or buys a pack of gems. A game that monetises entirely through subscriptions behaves differently. Revenue arrives through a funnel: install, trial start, then the first paid charge, often days later. If you want to know how to run UA campaigns for a game that monetises entirely through subscriptions, the answer starts with that funnel and with what you tell the ad network to chase.

This post covers UA strategies for games that monetise primarily through subscriptions: which event to optimise for, how trial length changes the picture, how to set a payback window, and how to judge channels fairly.

Why subscription games need a different UA setup

Ad networks learn from the events you send them. For an ad-funded game, every session produces a signal. For a subscription game, the signal that matters most is rare and slow. Only a small share of installs ever start a trial, and only some of those turn into paying subscribers.

That creates three problems:

  • Thin data. The deeper the event, the fewer conversions the network sees each day, so its model learns slowly.
  • Delay. A paid conversion that lands after a seven day trial reaches the network a week after the click. Campaigns are steered by old information.
  • Attribution limits. A purchase that falls outside the attribution window may never be credited to the ad that earned it.

None of this makes subscription UA impossible. It means you plan the funnel before you plan the budget.

Which event should you optimise for?

You have a ladder of events to choose from, each with a trade-off.

Install. The cheapest and fastest signal, and the weakest. The network finds people who tap install, not people who subscribe. It is useful for launch tests and creative screening, not for steady scaling.

Trial start. This is where most subscription games should begin. It gives the network far more conversions than a paid event, so it learns quickly and you can read results sooner. The risk is that trial-optimised traffic can include people who start a trial and never pay. You manage that risk by tracking trial-to-paid rate for every campaign and creative, not just the cost of a trial.

Paid conversion. Closest to revenue, and the right goal in principle. In practice it only works when paid conversions arrive quickly and in enough volume for the network to learn from. A good rule is to choose the deepest event that still gives each campaign enough conversions to judge. If a campaign sees only a handful of paid events a week, move one step up the ladder.

Many studios run a split: a trial-start campaign for volume and learning, and a smaller paid-event campaign to test whether the deeper signal improves quality once volume allows. Compare them on cost per paying subscriber, never on cost per trial alone.

How trial length changes your campaigns

Trial length is a product decision, but it shapes UA directly.

A short trial, or a paid first week, delivers the paid event fast. The network gets cleaner feedback and attribution windows are rarely a problem. A long trial pushes the paid event far from the click, which makes paid-event optimisation harder and slows your reads.

Longer trials often show higher trial-to-paid rates, but that does not prove that lengthening your trial will lift conversion. Test trial length as a separate experiment with your product team, and do not change it in the middle of a UA test, or you will not know which change moved the numbers.

Your attribution partner matters here too. Check how it handles delayed purchase events, which window it uses, and how renewals are reported. Our post on UA after IDFA and privacy changes explains why delayed and aggregated signals are now a fact of life on iOS, and why a measurement plan should come before the campaign.

Setting a payback window that matches subscription revenue

A subscription game earns over many billing periods, so day 7 ROAS looks poor even when the game is healthy. That does not mean you should ignore early numbers. It means you read them differently.

Build a simple chain from the funnel:

  1. Cost per install, from the network.
  2. Trial start rate, from your own data.
  3. Trial-to-paid rate, from your own data.
  4. Expected revenue per paying subscriber over a period you trust, built from cohorts that have actually renewed.

Multiply it through and you have a projected value per install that you can compare with what you pay. Our guide to LTV for mobile games shows how to build that cohort view, and the same method applies here with renewals in place of ad impressions.

Then choose a payback target you can live with. A long payback window ties up cash, so be honest about your runway. Studios with limited runway often favour annual plans and shorter first-charge paths for exactly this reason, because they bring revenue forward. For a fuller discussion of what return to aim for, see what is a good ROAS for mobile games.

Churn is the number that quietly breaks projections. Use observed renewal curves from real cohorts, split by channel, and refresh them often. A projection built on assumed renewal rates is a guess with a spreadsheet around it.

Judging channels and creatives fairly

Channels will not look alike. Applovin, Google Ads, Mintegral and Unity each reach players in different contexts, and the share of installs that start a trial can differ a lot between them. Judge each one on the same ladder: cost per trial, trial-to-paid rate, then cost per paying subscriber, and later on renewal behaviour.

Creatives deserve the same discipline. A creative that promises a free game may bring cheap installs and weak trials. One that shows the actual value of the subscription, what the player gets and how the progression works, usually brings fewer but better-qualified players. Test the message, not only the visual. Try a value-led hook against a gameplay-led hook, and check the plan or offer shown on the first paywall screen matches what the ad promised.

Some practical habits help:

  • Keep the path from ad to paywall as short as the game allows.
  • Test annual versus monthly plan prominence as a product experiment, then reflect the result in your creative.
  • Watch first-day retention by campaign, since trial users who never return rarely convert.
  • Retarget lapsed trial users carefully, and measure it as its own campaign.

Budgeting and testing in a slow-feedback model

Because feedback is slow, cash discipline matters more. Fix a test budget for each channel, run it long enough for trials to mature into paid or lapsed, and decide in advance what result earns more spend. Our UA budget planning guide covers how to split testing and scaling budget, and the same logic applies, with a longer wait before you read the outcome.

Scale in steps. Raise spend on the campaigns whose paying-subscriber cost holds up as volume grows, and cut the ones where cost per trial looks good but trial-to-paid fades.

Key takeaways

Subscription UA is a funnel problem before it is a bidding problem. Start with trial-start optimisation for volume, track trial-to-paid by campaign and creative, and move to paid-event optimisation only when the data supports it. Choose a payback window from renewal data you trust, and judge every channel on cost per paying subscriber rather than cost per install.

If you are planning acquisition for a subscription game and want a second opinion on your event setup, get in touch. We have run performance campaigns across Applovin, Google Ads, Mintegral and Unity, and we are happy to talk through your funnel.