How does lifetime value accumulate per cohort?
№ 137 · Cohort LTV curve
Definition
MetricCumulative revenue per install by days since install
UnitUSD cumulative per install; % of terminal LTV realised by day N
Cumulative LTV per cohort over days-since-install. F2P needs to compound; 60% of LTV in first 30 days means trouble (no long-term retention). Build by tracking per-user revenue across full lifecycle. Example: D30 LTV $2.20, D90 $3.80, D365 $7.40 (healthy compounding).
- LTV
- Lifetime value. Average gross profit per customer per period divided by that period churn rate, or summed over expected life.
- D30
- Day 30 retention. Share of a signup cohort returning on the thirtieth day after install or signup.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| D90 cumulative IAP revenue per install under $0.80, with the curve flat after day 30 | D90 cumulative IAP revenue per install $1.30 to $2.20; roughly 60% of D60 IAP revenue landed by day 7; D180 LTV of $1.80 to $2.50 casual and $4 to $6 midcore in Tier 1 | D90 cumulative IAP revenue per install above $2.50 in Tier 1 markets, with the curve still climbing between D90 and D180 |
Mobile free-to-play install cohorts, gross revenue before store fees unless stated, global with Tier 1 called out separately, 2024 to Q1 2026. LTV here is cumulative revenue per install, not per payer and not per daily active user. The terminal point is stated explicitly as D90 or D180 because an unqualified LTV number carries no information. · AppsFlyer publishes genre averages rather than percentiles, so the top30 and bottom30 bands are inferred from the spread between genre averages and are not a measured distribution. Two definitions collide constantly: IAP-only LTV against blended LTV including advertising revenue, which in casual titles is often the larger half. Advertising revenue also accumulates far faster than IAP, reaching 89% of its D60 total by day 7 against 60% for IAP, so a blended curve and an IAP curve have different shapes and cannot be read off the same axis. The D180 casual and midcore ranges are 2024 figures and are flagged as such.
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| Casino | not published as a percentile | D90 cumulative IAP revenue per install of $2.43 globally and $2.55 in North America and Europe | not published as a percentile |
| Midcore (strategy, RPG) | not published as a percentile | D90 $2.13 globally and $2.66 in North America and Europe; D180 LTV $4 to $6 in Tier 1 (2024 figure) | not published as a percentile |
| Casual | not published as a percentile | D90 $1.34 IAP plus $0.55 advertising; D180 LTV $1.80 to $2.50 in Tier 1 (2024 figure); D30 ROAS 47% on iOS and 15% on Android | not published as a percentile |
| Hypercasual | not published as a percentile | advertising-led, D90 advertising ARPU $0.22, with 63% of D60 advertising revenue landing on day 1 | not published as a percentile |
When it looks bad
Cohort curves stack in the wrong order, each newer cohort sitting below the one before it at the same day index, and every curve goes flat rather than continuing to climb after day 30.
D30 cumulative revenue per install of $0.94, $0.88 and $0.79 across three consecutive monthly cohorts, with each curve adding under $0.05 between D30 and D90.
What to do about it
- Plot cumulative revenue per install against the cohort's acquisition cost on the same axis so the payback date is read off the chart rather than calculated separately. Cost is fixed for a cohort on day zero, so the crossing point is the only date on the chart that carries a decision.
- Publish the curve separately for IAP and advertising revenue. Advertising reaches 89% of its D60 total by day 7 against 60% for IAP (AppsFlyer), so a blended curve that looks front-loaded may simply be ad-heavy, and extrapolating it past day 30 will overstate the terminal value.
- Cut the curve by media source, geography, platform and device tier before extrapolating anything. AppAgent's studio audits find teams routinely under-forecast long-term LTV because a single blended curve averages over cohorts whose shapes genuinely differ, and high-end device cohorts carry materially more IAP value.
- Refit the LTV model against current retention rather than 2022 or 2023 assumptions. D1 assumptions carried over from casual data of that period run several points high, the error compounds at D7 and again at D30, and the projected payback then never arrives.
Sources
- AppsFlyer D90 IAP ARPU of $2.43 casino, $2.13 midcore, $1.34 casual globally, rising to $2.66 midcore and $2.55 casino in North America and Europe. Advertising revenue reaches 89% of its D60 total by day 7 while IAP reaches 60% and subscriptions 52%. Day 1 alone is 57% of D60 advertising revenue and 38% of D60 revenue across all apps. Paid installs generate 59% of all IAP revenue in gaming. appsflyer.com ↗
- AppAgent In 2024, casual games in Tier 1 markets saw D180 LTV in the $1.80 to $2.50 range while midcore titles reached $4 to $6 or more. LTV curves differ materially by platform, geography, media source, campaign type and device tier, and teams routinely under-forecast long-term LTV by fitting a single blended curve. appagent.com ↗
- Liftoff and Singular Average D30 ROAS for casual games was 47% on iOS and 15% on Android. Sports games saw the highest D30 ROAS on iOS at 80%, followed by kids at 68% and strategy at 60%. RPG led on Android at 39% and tabletop was lowest at 7%. liftoff.ai ↗
- Moloco D7 ARPPU fell 3% year on year while D30 rose 3% and D90 rose 6%, and top advertisers improved D30 and D90 cohort LTV by up to 6% year on year, so value is accumulating later in the curve rather than earlier. moloco.com ↗
- GameAnalytics Median D7 retention of 3.42% to 3.94% and 75% of titles below 3% at D28, which sets the retention ceiling that any LTV curve is built on. gameanalytics.com ↗
- Sensor Tower (via GameDev Reports) Top 25 midcore games show D365 retention of 4% to 5% and declining, so the long tail that supports midcore LTV models is shortening. Average D7 among the top 25 casual games was 14.9% in December 2025. gamedevreports.substack.com ↗
- AppMagic By day 90 the share of App Store paying users who had spent more than $100 rose from 22% in 2024 to 32% in 2025 in hybridcasual, so payer depth is building earlier in the curve than it used to. gamedevreports.substack.com ↗
- Adjust Global CPI rose 30% to $0.56 and the median paid-to-organic install ratio rose 61% to 3.33, so the LTV curve has to clear a cost line that is moving up while the curve itself is flattening. adjust.com ↗
- AppsFlyer Among the fewer than 5% of users who purchase, transactions occur daily through roughly the first 10 days post install, which is where the steep section of the IAP curve comes from. appsflyer.com ↗
- Game Growth Advisor For IAP-dominant midcore and subscription-led games no primary source publishes a usable per-user revenue benchmark, so the curve should be modelled from the title's own payer share and average order value rather than borrowed. gamegrowthadvisor.com ↗