Most mobile gaming UA teams optimize the wrong window. They fixate on D1 and D7 because that’s what they’re compensated on. D30, D90, and payback period — the windows that actually determine whether the business survives — are someone else’s problem. Usually the product team’s. Sometimes the finance team’s. Often nobody’s, until cash gets tight.
According to Liftoff and Singular’s 2025 Casual Gaming Apps Report (analyzing 1.4 trillion impressions, 63 billion clicks, 2.5 billion installs, and $11.9 billion in ad spend between February 2024 and February 2025), casual games saw average D30 ROAS of 47% on iOS and 15% on Android in 2024. Same genre. Three times the return, purely from platform.
Numbers like that only mean something if you’re planning around payback period, not just LTV. A theoretical 365-day payback works only if you have the cash in the bank to fund that user pool for a year. Most gaming companies don’t. That’s the constraint that actually breaks UA budgets, not whether your CPI is $0.14 on Android or $1.41 on iOS.
The contrarian piece of this is the cheap CPI itself. UA teams hear “casual game CPI is $0.14 on Android” and assume cheap CPIs are a starting condition. They aren’t. The first 5, 10, 20 creatives you test won’t have cheap CPI. Cheap CPI is what you earn after creative velocity does its iteration work and the algorithm finds the audiences that convert. Treating cheap CPI as a planning assumption instead of an outcome is how teams end up scaling losing creatives because “the CPI is supposed to be cheap.”
Page Contents
The D1/D7 trap is structural, not analytical
The reason UA teams over-index on D1 and D7 is not that they don’t know D30 and D90 matter. It’s that the incentive structure inside most gaming companies stops at D7. The UA team is measured on installs, CPI, and short-window ROAS. The product team is measured on retention and monetization curves. Finance owns payback period. Three teams, three windows, no single owner of the metric that actually governs survival.
So when the UA team finds a creative that hits a great D7 ROAS, they scale it. They don’t have visibility into whether those D7 winners are actually D30 winners. And by the time the product team or finance team notices that the cohort is underperforming, the team has already burned a quarter’s worth of paid spend on it.
The fix isn’t analytical. The fix is to put D30 and payback period inside the UA team’s weekly review and into how the team is actually incentivized. A piece I wrote on Mobile Dev Memo back in 2019 argued that a chunk of Uber’s ad fraud problem traced back to exactly this kind of incentive misalignment — UA teams compensated on short-window metrics will rationally optimize for those metrics, even when the longer-window economics get worse. The teams that change the incentive consistently scale fewer creatives, but the ones they scale survive the full payback window.
What the 2025 Casual Gaming Apps Report actually shows
The Liftoff and Singular 2025 Casual Gaming Apps Report covers performance across 2.5 billion installs and $11.9 billion in ad spend between February 2024 and February 2025. The numbers worth internalizing:
| Genre | iOS CPI | Android CPI | Notes |
|---|---|---|---|
| Casual | $1.41 | $0.14 | 10x platform spread on the same genre |
| Casino | $21.03 | — | Highest iOS CPI in the report; payback math works because LTV is high |
| RPG | — | $4.29 | Highest Android CPI; long-window LTV does the lifting |
What we see: The genre-level CPI spread is real, but the platform spread on the same genre is bigger. A 10x CPI gap between iOS and Android casual games is the kind of gap that should be governing budget allocation, not nudged by it.
The D30 ROAS data tells the other half of the story:
| Genre | iOS D30 ROAS (2024) | Android D30 ROAS (2024) |
|---|---|---|
| Casual | 47% | 15% |
What we see: Same genre, 3x the D30 return on iOS. The cheap Android CPI does not translate proportionally into unit economics, because the monetization curve on Android casual is structurally weaker. This is the trap. If you optimize for Android CPI alone, you’ll end up over-invested in the platform with the worse payback math.
The report also flags an underused acquisition channel for casual games: non-gaming apps. 28% of casual game installs from non-gaming publishers come from utility and productivity apps, 25% from entertainment apps, and another 25% from photo and social media apps. If you’re a casual game and your spend is concentrated in gaming-native inventory (ad networks, Google UAC’s gaming placements), you’re leaving a structural acquisition channel on the table.
The cheap-CPI fallacy
Look at those genre-level CPI numbers again. The single most common mistake we see in mobile gaming UA: teams plan their budgets assuming a cheap aggregate CPI is the starting condition. It isn’t. It’s the outcome after months of iteration. A team that budgets for $0.14 CPI on Android casual from day one, gets $0.80 CPI for the first quarter, and concludes that the channel doesn’t work has not really tested the channel — it has exited the iteration loop too early.
This is also why “the CPI is supposed to be cheap” becomes a dangerous internal narrative. It rationalizes scaling losing creatives. The team sees a creative that’s converting at $0.50 Android CPI and thinks “well, it’s not $0.14, but it’s close enough.” It isn’t close enough. Aggregate CPIs at the lower bound exist at the tail of a long iteration distribution. The middle of the distribution is what most teams actually ship at, and the math on the middle does not work the way the tail does.
The fix is to plan budgets against the CPI we expect during the iteration phase, then scale past that only when the creative iteration loop has produced a winner that actually beats it. Cheap CPI is the destination, not the starting point.
Payback period vs LTV: the cash flow constraint
LTV is the number UA teams quote. Payback period is the number that actually governs whether you can keep buying users next month.
The math is simple. If your D365 LTV is $3.50 and your CPI is $1.41, your LTV/CPI ratio is 2.5x, which sounds healthy. But if your payback period is 365 days, then every dollar you spend on UA today is locked up for a year before it comes back. The constraint isn’t the ratio. The constraint is how much cash you can lock up before the bank account drops below what you need to keep paying salaries, vendors, and the rest of the company.
This is the conversation that doesn’t happen between UA, product, and finance teams. The UA team says “the unit economics work.” Finance says “we don’t have the cash to fund 365-day payback at this scale.” Both are right. The actual question is: what’s the largest UA budget we can run such that the cumulative cash lockup doesn’t exceed our working capital?
That number is almost always smaller than the LTV/CPI ratio suggests is possible. Teams that get this right run UA at a lower spend ceiling but at higher payback velocity. Teams that get it wrong scale UA past the cash flow constraint and then have to slam the brakes when finance notices.
Where creative velocity actually moves the numbers
The lever that compresses payback period faster than anything else is creative velocity. Not creative quality in isolation. Velocity.
Here’s why. The first 5, 10, 20 creatives you test on any new account will not have cheap CPI. They can’t. The algorithm hasn’t found the audiences that convert yet, the creative hasn’t found the hooks that work, and the team hasn’t found the message that lands. Cheap CPI is the output of an iteration loop, not an input to it.
Teams that ship 3 new creatives a month spend 3 to 6 months in the expensive-CPI phase, because they’re getting only 3 data points a month. Teams that ship 20 new creatives a month compress that learning into 4 to 6 weeks. The right monthly creative volume by budget is the planning number to start from. The expensive-CPI phase is real for both teams. The difference is how fast they exit it.
This is also why “AI will give you cheap creative production” is the wrong framing. The cost of producing the creative is not the constraint. The constraint is the speed at which you can learn what works. AI helps if it lets you ship more distinct hook angles per week. It doesn’t help if it just produces more variants of the same hook faster.
What this means for you
- Tie incentives to D30 and payback period, not just D7. If the UA team is only compensated on short-window metrics, they will rationally optimize for those, and the longer-window economics will suffer. The fix has to be in the incentive structure, not just in adding new dashboards.
- Plan UA spend against the cash lockup constraint, not the LTV/CPI ratio. The ratio tells you if the unit economics are positive. The constraint tells you how much you can spend before you run out of working capital.
- Budget against iteration-phase CPI, not the headline number you’d hope to land at. Cheap CPI is what you earn after iteration. It’s not a starting condition.
- Maximize creative velocity, not creative cost efficiency. Shipping 20 distinct hook angles a month compresses the expensive-CPI phase faster than shipping 3 polished concepts. The cost of the iteration loop is much smaller than the cost of staying in the expensive-CPI phase for an extra quarter.
- Look at non-gaming inventory for casual games. Liftoff and Singular’s data shows utility, entertainment, and photo/social apps drive over 75% of casual game installs from non-gaming publishers. If your spend mix is gaming-native only, you’re missing a structural acquisition channel.
Need help structuring your mobile gaming UA budget around payback math instead of CPI? Talk to RocketShip HQ about how we apply these frameworks for apps spending $50K+/month on UA.
Frequently Asked Questions
Why do mobile gaming UA teams over-index on D1 and D7 instead of D30?
Structural, not analytical. UA teams are typically compensated on installs, CPI, and short-window ROAS — metrics that are observable in the first week. D30, D90, and payback period sit with product and finance teams, so there’s no single owner of the metric that actually governs survival. The fix is to put D30 and payback period inside the UA team’s weekly review and into how the team is incentivized. When teams are paid on short-window metrics, the longer-window economics predictably suffer — the fix has to be in the incentive, not just the dashboard.
What does the 2025 Casual Gaming Apps Report say about iOS vs Android CPI?
Casual games saw average CPI of $1.41 on iOS and $0.14 on Android between February 2024 and February 2025. The same report shows D30 ROAS in 2024 of 47% on iOS and 15% on Android for casual games. The 10x CPI gap is real, but so is the 3x ROAS gap. Cheap Android CPI does not translate proportionally into unit economics because the monetization curve on Android casual is structurally weaker than iOS.
Why is payback period more important than LTV for mobile gaming UA?
LTV tells you if the unit economics are positive. Payback period tells you how long your money is locked up before it comes back. A theoretical $3.50 D365 LTV at $1.41 CPI looks like a healthy 2.5x ratio, but if the payback period is 365 days, every dollar spent today is locked up for a year. The real constraint on UA budget is how much cash you can lock up before your working capital drops below what you need to run the rest of the company. Teams that get this right run UA at a lower spend ceiling but higher payback velocity.
Is cheap CPI a starting condition or an outcome?
An outcome. The first 5, 10, 20 creatives a team ships on any new account will not hit the lower-bound CPI numbers that get cited as genre benchmarks. The algorithm hasn’t found the converting audiences yet, the creative hasn’t found the working hooks, and the team hasn’t found the message that lands. Those lower-bound CPIs exist at the tail of an iteration distribution. The middle of the distribution is what most teams actually ship at, and the math on the middle does not work the way the tail does.
How does creative velocity compress payback period?
Faster iteration finds the converting creative-audience combinations faster, which shortens the expensive-CPI phase that every new account or new creative cycle goes through. Teams that ship 3 new creatives a month spend 3 to 6 months in the expensive-CPI phase. Teams that ship 20 new creatives a month compress that learning into 4 to 6 weeks. The cost of running the iteration loop is much smaller than the cost of staying in the expensive-CPI phase for an extra quarter.
What’s the right CPI assumption for budgeting a new mobile game launch?
The CPI you expect to see during the creative iteration phase, not the lower-bound CPI you’d hope to land at after months of iteration. Lower-bound CPIs exist at the tail of the iteration distribution. Budgeting against them commits you to spend levels that only work if you actually reach that tail. Budget against the iteration-phase CPI for your genre and geo, then scale past it only when the iteration loop has produced a creative that beats it consistently.
Where else should casual games look for installs beyond gaming inventory?
Per the 2025 Casual Gaming Apps Report, 28% of casual game installs from non-gaming publishers come from utility and productivity apps, 25% from entertainment apps, and 25% from photo and social media apps. That’s over 75% of non-gaming-sourced installs concentrated in three publisher categories. Casual games with spend mixes concentrated in gaming-native inventory are missing a structural acquisition channel that real data confirms is meaningful.
Why does casino have a $21+ iOS CPI in the Liftoff data, and how does that work?
Casino monetization curves are aggressive. LTV from a paying casino user is high enough that the genre can absorb iOS CPIs in the $20s and still hit positive payback within a reasonable window. The genre-level CPI difference between casino and casual isn’t a failure of casino UA — it’s a feature of how casino LTV works. The right question for any genre isn’t “is the CPI cheap?” It’s “does the payback math work at this CPI?”
Looking to scale your mobile gaming UA with frameworks that respect the cash flow constraint? Talk to RocketShip HQ to learn how our payback-period planning works for gaming apps spending $50K+/month.
Not ready yet? Get strategies and tips from the leading edge of mobile growth in a generative AI world: subscribe to our newsletter.
Related Reading
