The short answer
The mobile game KPIs that matter most are retention (do players come back on day 1, 7 and 30), monetisation per user (ARPDAU and LTV), and acquisition efficiency (CPI and ROAS). A game is commercially healthy when the lifetime value of a player comfortably exceeds the cost of acquiring one. Everything else, from session length to payer conversion, helps explain why those numbers move.
The formulas
| KPI | What it means | Formula |
|---|---|---|
| D1 / D7 / D30 retention | Share of a day's new players who return on day 1, 7 or 30 | Players from the install cohort active on day N ÷ players in the cohort |
| DAU / MAU | Daily and monthly active users | Unique players active in a day or a month |
| Stickiness | How often monthly players play | DAU ÷ MAU |
| Session length / sessions per day | Engagement depth and habit | Total playtime ÷ sessions; sessions ÷ DAU |
| Payer conversion | Share of players who buy | Paying users ÷ active users |
| ARPDAU | Average revenue per daily active user | Daily revenue (ads + purchases) ÷ DAU |
| ARPPU | Average revenue per paying user | Purchase revenue ÷ paying users |
| ARPU | Average revenue per user over a period | Revenue ÷ active users in the period |
| LTV | Revenue a player generates over their lifetime | Cumulative revenue from a cohort ÷ installs in the cohort (by day 7, 30, 90...) |
| CPI | Cost per install | Ad spend ÷ installs from that spend |
| ROAS | Return on ad spend | Revenue from a cohort ÷ spend to acquire it (often by day 7 or 30) |
| Churn | Players who stop playing | 1 − retention for the period |
Retention comes first
Retention is measured in cohorts: everyone who installed on a given day is followed over time. Mixing cohorts, such as comparing today's actives with yesterday's installs, produces numbers that look better than reality. Day-one retention mostly reflects the first session and tutorial; day seven reflects whether there is a reason to return; day 30 reflects depth and long-term goals.
For context, GameAnalytics' 2026 benchmarks, covering more than 16,000 mobile games with at least 1,000 monthly users in 2025, report:
| Metric | Median game | Top 25% | Top 10% |
|---|---|---|---|
| Day 1 retention | About 22% | Just above 30% | About 40% |
| Day 7 retention | Just under 4% | About 6–7% | About 11–12% |
| Day 30 retention | Under 1% |
The same report put median daily playtime at about 12 minutes. GameAnalytics stresses that genres and business models produce different curves, and calls the familiar 40/20/10 rule (40% day 1, 20% day 7, 10% day 30) a reference point rather than a pass mark. Compare against your genre and your own previous builds.
Monetisation per user
ARPDAU tells you what an average day of play earns, combining ads and purchases. ARPPU and payer conversion separate how many people pay from how much they spend; a game can grow revenue by converting more players or by giving payers more worth buying. Count ad revenue properly: in ad-supported games it is often most of the income, and leaving it out makes LTV look far too low. Our guide to ads vs in-app purchases explains the trade-offs.
LTV, CPI and ROAS: the business test
Lifetime value is the number that decides how much you can spend on marketing. Teams usually track LTV at day 7, 30 and 90 for each cohort and project it forward. CPI is what installs cost in a given market and channel. ROAS compares a cohort's revenue with what it cost to acquire, often measured at day 7 and day 30. If projected LTV is below CPI, buying installs loses money, however good the game feels.
Instrument from the first build
GameAnalytics recommends a minimum event set of session start, tutorial complete, level complete, purchase and ad view. Add funnel steps for your first session, economy sources and sinks, and the points where players quit. Firebase's Google Analytics is free and widely used with Unity; whatever tool you use, agree event names before development so data stays comparable between builds.
Mistakes that mislead
- Mixed cohorts and comparing players acquired from different channels as if they were the same.
- Ignoring ad revenue in LTV and ARPDAU.
- Over-reading small tests. A few percentage points on a few hundred players is noise.
- Celebrating averages. A rise in ARPPU can hide a fall in payers.
- Old benchmarks. Retention across the market has fallen in recent years; use current, genre-specific data.
We set up analytics, dashboards and LiveOps for the games we build and for live games that need it. See LiveOps outsourcing and our soft launch guide for how these numbers drive launch decisions.
Sources: GameAnalytics: 2026 mobile and PC benchmarks · GameAnalytics: how to think about retention · GameAnalytics: post-launch analytics · Firebase: Google Analytics. Checked October 14, 2026.
Related: Ads vs IAP · LiveOps outsourcing · Mobile game development
Originally published on A Plus Games Studio, a game, AR and VR development studio in Sacramento, CA.
Top comments (0)