How sticky are new users over the first 90 days?
№ 069 · Retention curves
Definition
MetricClassic (day-exact) retention at D1 / D7 / D30
Unit% of install cohort active on day N
Two lines (old cohort vs recent cohort) showing % active at D1, D7, D30, D60, D90. Read whether recent improved over old. Build by counting active users per day-since-install, grouped by install cohort. Step changes follow product launches or onboarding fixes. Example: Old cohort D30 = 14%; Recent D30 = 20%.
- D1
- Day 1 retention. Share of a signup cohort returning the day after install or signup.
- D7
- Day 7 retention. Share of a signup cohort returning on the seventh day after install or signup.
- D30
- Day 30 retention. Share of a signup cohort returning on the thirtieth day after install or signup.
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| D1 under 15%, D7 under 6%, D30 under 3% | D1 25-26%, D7 11-13%, D30 5-7% | D1 35%+, D7 18%+, D30 10%+ (subscription apps reach D30 ~14%) |
All-vertical consumer mobile apps, global, classic retention on install cohorts, 2024-2026 network data (Adjust, AppsFlyer aggregates, GameAnalytics). Week-over-week retention past D30 typically stabilizes at 2-4%; D90 is rarely published, so treat any D90 target as internal. · Classic vs rolling retention changes the number materially; most published benchmarks are classic. Pooled medians are dragged down by the long tail of small games. Top30 is converted from sources that publish 75th percentile or 'strong performer' figures, so it approximates better-than-70th-percentile. iOS retains structurally better than Android (D30 8% vs 6%, Adjust).
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| Mobile games (pooled, all sizes) | D1 10-11.5%, D7 ~1.5% | D7 3.4-3.9%, D28 under 3% for most | D1 27%+, D7 7-8% |
| Subscription apps | D30 under 8% | D30 ~14% | D30 18%+ |
| Marketplace / commerce apps | D30 under 4% | D1 25%, D7 15%, D30 8% | D1 30%+ |
When it looks bad
Cohort lines that never flatten: each curve still falling at D30-D90 with no plateau, and newer cohorts stacked below older ones.
D1 18%, D7 5%, D30 1.5% and still declining, with the May cohort tracking 30% below the January cohort at every point.
What to do about it
- Rebuild the first session around one core value action; apps that get users to a meaningful first action in session one see 2-3x better D7 retention (Appcues 2026 aggregate).
- Trigger re-engagement inside the 3-7 day dormancy window; early automated triggers return lapsing users at 2-3x the rate of waiting a week or more (Appcues).
- Front-load value on day zero; the majority of trial cancellations happen on day zero, so the first session has to demonstrate the payoff, not promise it (RevenueCat 2026).
- Split the curve by platform before acting; iOS D30 runs 8% vs 6% on Android (Adjust), so a mix shift toward Android can read as product decay when it is acquisition mix.
Sources
- Adjust Global: D1 26%, D7 13%, D14 10%, D30 7%; iOS D1 27% / D30 8%, Android D1 24% / D30 6% adjust.com ↗
- Adjust Same global curve; an app is considered high-performing if it retains over a third of users after install adjust.com ↗
- GameAnalytics Games: top-quartile D1 26.5-27.7% (31-33% on iOS), bottom-quartile D1 10-11.5%; median D7 3.4-3.9%, top quartile 7-8%; 75% of games sit below 3% D28 gameanalytics.com ↗
- UXCam A typical app loses 75% of users within three days of install; category tables reflect ~75th percentile performers; retention past D30 stabilizes at 2-4% week over week uxcam.com ↗
- Phiture Cross-industry 2025-2026 averages: D1 25-26%, D7 11-13%, D30 5-7% phiture.com ↗
- Digital Applied Cross-category averages D1 26%, D7 11%, D30 5.4%; subscription apps D30 ~14% vs 5.4% for ad-supported (citing AppsFlyer) digitalapplied.com ↗
- RevenueCat Year-one subscriber retention 27-28% for annual plans; the first session (day zero) is where users decide both whether to pay and whether to stay revenuecat.com ↗
- Stream (citing Adjust Mobile App Trends 2025) Marketplace apps: D1 25%, D7 15%, D30 8% in 2024, all down 1-2 points vs 2023 getstream.io ↗