ROAS and LTV for Mobile Games: How to Tell Whether a Campaign Pays Back

Every user acquisition campaign answers one question in the end: did the players it brought in earn back what it cost to bring them in? Two numbers answer it. Lifetime value (LTV) is what a player is worth over their time in the game. Return on ad spend (ROAS) is how much revenue a campaign has returned so far, relative to what it cost. LTV is the destination; ROAS is how far you have travelled towards it.

Both are simple to define and easy to misread. This guide covers how to calculate them, how to read them by cohort and by day, and how to set a target that tells you early whether a campaign will pay back.

How to calculate ROAS

ROAS is revenue divided by spend, usually shown as a percentage.

Two details make the number meaningful. First, revenue should be what the game actually keeps — after store fees and, for ad revenue, as reported by your mediation platform — or you will overstate the return. Second, the revenue must belong to the same users the spend acquired, which is why ROAS is always measured by cohort.

Cohort ROAS and the day number

A cohort is the group of users who installed on the same day, or in the same week, from the same campaign. Cohort ROAS tracks the revenue that group produces as it ages: D0 ROAS is revenue on install day, D7 ROAS is the cumulative revenue by day seven, D30 by day thirty, and so on.

An illustration, with invented numbers: a campaign spends $1,000 to acquire a cohort. By day seven those players have produced $300 in purchases and ad revenue. D7 ROAS is 30%. By day thirty they have produced $650, so D30 ROAS is 65%. The campaign has not paid back yet — but whether it will depends on how the curve keeps rising, which is what a target is for.

Always compare cohorts at the same age. A campaign launched last week will always show a lower cumulative ROAS than one launched two months ago, and reading them side by side without aligning the day is one of the most common mistakes in UA reporting.

How to calculate LTV

LTV is the total revenue an average player in a cohort produces over their lifetime in the game. You can only know it exactly in hindsight, so in practice teams work with LTV at a horizon — D30 LTV, D90 LTV, D180 LTV — and with predictions.

• Observed LTV at day N: cumulative cohort revenue by day N ÷ number of installs in the cohort.
• Simple model: average revenue per daily active user × average lifetime in days. Quick, but it hides the shape of retention.
• Retention-curve model: sum, day by day, the share of the cohort still active × revenue per active user on that day. Closer to reality, and the basis of most forecasts.

Note the link between the two numbers: ROAS at day N equals LTV at day N divided by the cost per install. That is why the cheapest way to think about a campaign is often as LTV against CPI — and why both lowering CPI and improving the quality of the players you acquire move ROAS.

Payback period

The payback period is how long a cohort takes to reach 100% ROAS. It matters as much as whether it gets there, because money spent today and recovered in a year is tied up for that year. The right payback window depends on cash flow, the game's monetization model and how confident you are in long-term retention.

• Ad-led games tend to earn early and flatten quickly, so payback needs to come early or not at all.
• IAP-led games often earn slowly and keep earning, so a longer payback can still be a good investment — if the retention holds.
• Hybrid games sit between, and their curve depends on how many players convert to payers and when.

Setting an early ROAS target

Waiting until day 180 to decide whether a campaign works is not an option, so teams work backwards from the payback goal to an early checkpoint, usually D7.

1. Take mature cohorts from your own game and calculate their ROAS at day 7 and at your payback day.
2. Divide one by the other to get a multiplier: how much the cohort's revenue typically grows between the two.
3. Divide your payback goal by that multiplier to get the D7 target.
4. Judge new campaigns against that target once they reach day seven, and re-check the multiplier as more cohorts mature.

Another illustration: if your mature cohorts show that D180 revenue is typically about three times D7 revenue, and you want to break even by D180, a campaign needs roughly 33% D7 ROAS to be on track. The multiplier is specific to your game — never borrow another title's.

Predicting LTV from early data

Prediction fits a curve to the first days of a cohort and extends it. The inputs that carry most of the signal are early retention, early payer conversion, early ad engagement and the in-app events that correlate with long-term value in your game — finishing the tutorial, reaching a level, making a first purchase.

Keep predictions honest by comparing every forecast against what the cohort actually did once it matured, and by treating forecasts for small cohorts with caution: a handful of large spenders can make a tiny cohort look exceptional.

Measurement limits

Attribution on iOS has become less precise since App Tracking Transparency and SKAdNetwork: user-level data is limited for users who do not opt in, and conversion values arrive delayed and coarse. ROAS by campaign on iOS is therefore often modelled rather than counted. SKAdNetwork and creative measurement explains what can still be measured per creative.

Mistakes that flatter ROAS

• Gross revenue instead of net. Store fees and refunds belong out of the number.
• Ignoring ad revenue. For hybrid games, purchases alone understate the return.
• Comparing cohorts at different ages.
• Judging on CPI alone. Cheap installs that leave do not pay back.
• Trusting small cohorts. One whale is not a trend.
• Including organic installs in a paid cohort, which inflates paid ROAS.

Where playable creative fits

Creative affects both halves of the ROAS equation. It sets cost, through the install rate that decides how cheaply a network can deliver; and it sets value, through who it persuades to install. A creative that overpromises can produce a low CPI and a poor ROAS, because the players it attracts were expecting a different game.

Playables are useful here because the viewer has already played the loop before installing. Test them by cohort, not only by CPI — compare D7 ROAS for players acquired by a playable against those acquired by your best video. Playable Ads Maker makes the variants cheap to produce: build from a game template or from scratch, export for every major network, and let the cohort data decide. The A/B testing guide covers how to structure the test.