What's the shape of engagement intensity?
№ 077 · Session frequency distribution
Definition
MetricDistribution of daily (or weekly) session counts per active user
Unitsessions per active user per day; histogram of users by session count
Histogram of sessions per user per week. Usually bimodal (casuals and core users). Build from session event logs. The right peak is where retention and revenue live. Example: 38% users 1-2 sessions/week, 22% 3-7, 12% 8+ sessions (the power users).
Benchmarks
| Bottom 30% | Median | Top 30% |
|---|---|---|
| Mass piled at exactly one session per day, typically a notification open, with sessions per DAU near 1.0-1.2 | Roughly 2 sessions per active user per day, stable from day 1 to day 30 on a per-active basis; per-install session counts decay steeply (about 1.6 on day 0 to ~0.2 by day 30) | A distribution with a second mode at 3-4+ daily sessions and a meaningful tail (games average 4 sessions per day, midcore 6-7; social apps carry the highest counts) |
Consumer apps globally, sessions per user per day, 2021-2025 Adjust network data and GameAnalytics 2025 for games. Denominators differ by source: per-install counts (which decay with cohort age) vs per-active-user counts (which are stable); this card should be built per active user. · Full distribution shapes are rarely published; public data gives averages, so the shape guidance is convention plus adjacent data. The Adjust ~2 sessions figure is 2021-2022 data, flagged as dated. Push-triggered opens counted as sessions inflate the 1-session bar and are the main measurement artifact to strip.
| Category | Bottom 30% | Median | Top 30% |
|---|---|---|---|
| Mobile games | under 2 | 4 sessions per day | 6-7 sessions per day (midcore) |
| Social / content | single daily push-driven open | above the ~2 all-app average | highest of any vertical, multiple daily check-ins |
| Utility / transactional | 1.0, single-purpose opens only | 1.2-1.5 per DAU | 2+ per DAU with high task completion |
When it looks bad
A single dominant bar at one session per day driven by a notification open, with no second organic mode and a session length under a minute.
70% of DAU log exactly one push-driven session under 60 seconds; sessions per DAU sits at 1.1 against a ~2.0 norm.
What to do about it
- Split notification-triggered sessions from organic before optimizing anything; counting push opens as engagement hides the fact that the product has one artificial visit, not a habit.
- Build a second daily use case with its own trigger (a morning check plus an evening ritual) so the mode of the distribution moves from 1 toward 2.
- Trigger re-engagement off session interval, not calendar rules: message a user when their gap exceeds their own median interval, which catches decay per user rather than per cohort.
- For ad-monetized surfaces, prefer more short sessions over one long one, since impressions refresh per session; cap interstitial frequency so the extra sessions do not cost retention.
Sources
- GameAnalytics Players average 4 sessions per day; midcore 6-7; median session length 5-6 minutes, top quartile 8-9 gameanalytics.com ↗
- Adjust Sessions per user per day ~2.05-2.07 on day 1 and holding near 2 at day 30 on a per-active basis (dated, flagged) adjust.com ↗
- Adjust Per-install decay: North America 1.66 sessions per day on day 0 falling to 0.18 by day 7 and 0.09 by day 30 adjust.com ↗
- Business of Apps Average daily sessions drop from 1.6 to 0.22 after day one; social apps have the highest daily session counts, food and drink the lowest businessofapps.com ↗
- Coinis (citing Adjust and AppsFlyer 2024 benchmarks) Session lengths by vertical: gaming 8-12 min, social 4-7, shopping 2-4, utility 1-3; MMP vs analytics session counts diverge 5-20% coinis.com ↗
- CFOmatrix 2+ sessions per DAU is a strong signal for gaming and social; 1.2-1.5 acceptable for high-intent utility apps cfomatrix.in ↗