When does each subscriber cohort pay back acquisition cost, and how does retained revenue compound month after month?
№ 089 · Subscriber ROAS curve
Definition
MetricCumulative subscription revenue recovered against acquisition spend by cohort age
Unit% of acquisition spend recovered at D30/D180/D365; months to 100% payback
Cumulative gross margin per acquired subscriber divided by CAC across months. Multiple lines for cohorts or channels. Build by tracking per-cohort revenue minus content allocation against CAC monthly. Example: Search subs 1.0x at M5; Display subs 1.0x at M16 (search far better).
- ROAS
- Return on ad spend. Revenue attributed to a campaign divided by the spend on that campaign.
- CAC
- Customer acquisition cost. Fully loaded sales and marketing spend in a period divided by new customers won in that period.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| Curve flattening below 100% before month 12; D365 ROAS under 80% with newer cohorts starting lower than older ones | Payback around 9-12 months; D365 ROAS roughly 100-120% on blended paid channels | Full CAC payback within ~6 months; D365 ROAS at 150%+ with the curve still climbing from renewals |
Consumer subscription media apps, blended paid acquisition, subscription revenue only (no ad-revenue layer), 2024-2026 network data. ROAS defined as cumulative net revenue over acquisition spend for the cohort; percentile cut points are a synthesis, since networks publish revenue curves rather than payback distributions. · This is subscriber ROAS in a subscription media business: cohort revenue compounds through renewals, unlike one-shot ecommerce ROAS. Published anchors: hard-paywall apps generate $2.32 revenue per install by D14 and $3.09 by D60 vs $0.27/$0.38 for freemium (RevenueCat 2026); converted trial users produce ~50% of D90 subscription revenue (AppsFlyer); iOS subscription ARPU reaches $8.39 by D90 vs $1.54 on Android (AppsFlyer). Payback targets shift with plan mix, since annual plans collect cash up front.
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| iOS subscription apps | not published | $8.39 D90 ARPU per user (AppsFlyer); revenue-per-install materially above Android in RevenueCat store-level data | not published |
| Android subscription apps | not published | $1.54 D90 ARPU per user, roughly one fifth of iOS (AppsFlyer), directionally corroborated by RevenueCat store splits | not published |
When it looks bad
Cohort curves flattening below the 100% line before month 12, with each newer cohort starting lower and bending earlier than the one before it while spend is held constant.
D30 ROAS 35%, D180 70%, curve flat at 85% by month 10 across the last three quarterly cohorts.
What to do about it
- Judge channels on month 2+ slope, not D7 ROAS: set the evaluation window to your payback target as Liftoff prescribes, because a channel with weak D7 and strong D30-D90 renewals beats a front-loaded one in a renewal business
- Test a hard paywall or intro price against freemium: the D60 revenue-per-install gap is 8x ($3.09 vs $0.38, RevenueCat 2026), and intro pricing also cleans the conversion signal ad networks optimize on
- Shorten or restructure trials: converted trial users are ~50% of D90 revenue (AppsFlyer), so testing 3-7 day trials against the media-typical 30 days pulls the whole curve left
- Promote annual plans at the first paywall; the converting share pays back on day 0 and annual cohorts retain ~44% at month 12 (RevenueCat), steepening both the front and the tail of the curve
Sources
- RevenueCat Hard paywall revenue per install $2.32 at D14 and $3.09 at D60 vs freemium $0.27/$0.38, an 8x D60 gap; discounting warns directly against killing the payback period revenuecat.com ↗
- AppsFlyer Subscription apps: iOS D90 ARPU $8.39 vs Android $1.54; converted trial users generate 50% of D90 revenue; organics 65% of subscription revenue appsflyer.com ↗
- AppsFlyer IAA apps reach D90 ROAS of 95% (Android) and 80% (iOS), a contrast curve showing ad-monetized revenue front-loads while subscription revenue compounds appsflyer.com ↗
- RevenueCat Developers keep trials short to keep CAC payback short; media and entertainment apps run longer trials, up to 30 days, lengthening payback revenuecat.com ↗
- Liftoff ROAS targets must match the payback window of the business model; the same D7 or D28 ROAS reading can be healthy or a red flag depending on the target payback horizon liftoff.ai ↗
- Digital Applied Subscription apps retain ~14% of users at D30 vs 5.4% cross-category, the retention base that makes the ROAS curve compound (Tier 3 corroborator only) digitalapplied.com ↗