Unit economics
Mobile app marketing budget: how to calculate and allocate it
There is no useful market average that answers "how much budget do we need?" The answer depends on allowable CPA, the speed at which you need evidence, the number of markets, and how many hypotheses the team wants to test.
Three numbers you need before setting a budget
Start with allowable CPA, the number of conversions needed, and the time available to collect them. If a subscription can support $30 in acquisition cost and the team needs 100 purchases for an initial read, the media portion of the test is already about $3,000. Creative production, measurement, and failed hypotheses sit on top of that.
Base allowable CPA on contribution margin LTV, not gross revenue. Subtract store fees, refunds, taxes, and variable costs. The less you know about retention, the more conservative the forecast should be.
A test budget should buy information
Early budget is not there to maximize install volume. It should answer a specific question. Can a new UGC angle produce trials below $12? Does a web-to-app funnel convert better than a direct store path?
One campaign, one ad, and a few days rarely provide a dependable answer. Plan several independent hypotheses and a minimum event count for each. If the purchase event is too rare, optimize for an earlier event only when it has a proven relationship with payment.
A practical monthly allocation
| Area | Starting range | What it covers |
|---|---|---|
| Proven campaigns | 50-65% | Combinations with validated economics |
| New hypotheses | 20-30% | Creative, audiences, markets, and funnels |
| Production | 10-20% | UGC, editing, static assets, and adaptations |
| Reserve | 5-10% | Seasonality, bid inflation, and quick opportunities |
These ranges are not universal. A new app will spend more on testing. A mature product may send more into scale, but creative production cannot drop to zero. Paid social loses momentum when the account runs out of new messages.
Why a tiny budget can be expensive
An underfunded test often ends without a decision. The company spends money but collects too few events to separate chance from signal. The test then has to be repeated, or the team acts on weak evidence.
Google recommends relating daily budget to the target action cost. Its guidance includes a starting point of 50 target CPIs or 10 target CPAs per day for some App campaign setups. This is not a performance guarantee. It shows how much room an automated strategy may need.
When to increase spend
Raise the budget when economics hold across several cohorts, event delivery is reliable, and the account does not depend on a single winning ad. Make changes large enough to create new volume, but do not change budget, bids, geography, and the optimization event at the same time.
Pause when CPA rises without better user quality, the platform shifts spend into weak inventory, or post-install conversion falls. The right move may be to stop buying traffic until the team ships new creative or repairs onboarding.
Build a budget model the team can operate
Track weekly spend, installs, target events, purchases, revenue, and expected LTV. Mark delayed conversions separately. Keep channel, country, platform, and creative hypothesis attached to every row.
The budget then becomes a portfolio of decisions. Money moves from weak hypotheses to strong ones, while the team can see whether the real problem is traffic cost, store conversion, activation, or monetization.