Analytics tools will happily give you hundreds of charts. Most of them will not help you make a single decision. The mobile app metrics that matter are the few that tell you whether people get value from your app, whether they come back, whether the app is stable and whether the business model works. This article explains which ones to track, how to read them and how to set up measurement properly from the start.
Beware the vanity metrics
Total downloads is the classic vanity metric: it only ever goes up, and it says nothing about whether anyone still uses the app. Registered users, total sessions and page views have the same problem. They look good in presentations but cannot tell you what to fix. A useful metric changes in response to your decisions and points to an action.
A framework for mobile app metrics: follow the user journey
Organise metrics by the stages a user passes through. For each stage, pick one or two numbers that matter for your product.
| Stage | Question | Example metrics |
|---|---|---|
| Acquisition | Are the right people finding and installing it? | Installs by source, store listing conversion rate, cost per install |
| Activation | Do new users reach the first moment of value? | Onboarding completion, time to first key action |
| Engagement | Do they use the core features? | Active users, feature adoption, key actions per user |
| Retention | Do they come back? | Cohort retention by day, week or month |
| Revenue | Does usage turn into value for the business? | Conversion to paid, revenue per user, churn |
| Quality | Does the app work reliably? | Crash-free users, app-not-responding rate, startup time, ratings |
Activation: the most underrated mobile app metric
Define the activation event: the action that shows a new user has experienced what the app is for. For a delivery app it might be placing a first order; for a field tool, completing a first job report; for a budgeting app, connecting an account and seeing a first summary.
Then measure what share of new users reach it, and how long it takes. Activation problems are often cheaper to fix than acquisition problems: shortening sign-up, removing a confusing step or improving an empty state can lift the value of every install you already pay for.
Retention: read it as cohorts
Retention shows whether people return. Measure it in cohorts: group users by the day or week they installed, then track what fraction is active on day 1, day 7, day 30 and so on. Plotted, cohorts form retention curves.
- A curve that keeps falling towards zero means users are not finding lasting value.
- A curve that flattens means you have found a group that sticks. The level where it flattens matters more than the initial drop.
- Comparing cohorts shows whether product changes are working: are newer cohorts retaining better than older ones?
Match the retention window to natural usage. A messaging app should be judged on daily use; a tax app, an insurance app or a travel booking app may be used only a few times a year, and judging it on daily activity would be misleading.
Engagement: depth, not just activity
Daily active users (DAU) and monthly active users (MAU) count unique users in a period. The ratio DAU/MAU, sometimes called stickiness, indicates how habitual use is. Define "active" carefully: opening the app is weaker evidence than completing a meaningful action.
Track feature adoption too: what share of active users use each core feature. Features almost nobody uses are candidates for redesign or removal, which simplifies the app for everyone.
Revenue metrics
- Conversion rate from free to paid, or from browsing to purchase.
- Average revenue per user, in total and per paying user.
- Churn: the rate at which subscribers cancel.
- Lifetime value: an estimate of revenue per user over their whole relationship with you. Compare it with acquisition cost to see whether growth is sustainable. Early estimates are rough; treat them with caution.
For internal business apps, "revenue" becomes operational value: jobs completed per technician, time from visit to invoice, or reduction in paperwork errors.
Quality metrics
Instability destroys retention quietly. Track:
- Crash-free users and sessions, from a crash reporting tool, broken down by app version, OS version and device.
- App not responding (ANR) rate on Android, where the app freezes. Google Play Console's Android vitals reports this, and Google has said poor vitals can affect an app's visibility on Play.
- Startup time and the speed of key screens.
- API error rates and latency from your backend monitoring, since many "app" problems start on the server.
- Ratings and review themes, read regularly, not just averaged.
Setting up measurement properly
- Write an event plan before development: a list of events (such as
order_placed), their properties (such as order value and payment method) and when they fire. Consistent naming prevents a mess of duplicate events later. - Track meaningful actions, not every tap.
- Tie events to releases, so you can compare behaviour before and after a change.
- Respect privacy. Collect only what you need, avoid personal data in event properties, honour consent requirements in your markets, and keep your app store privacy declarations consistent with what your analytics SDKs collect.
- Verify the data. Test events in debug builds and periodically reconcile analytics with backend records, such as orders in the database.
Server-side data is often the most reliable source for business metrics, because it is unaffected by blocked trackers or users who decline analytics. A sound database and reporting setup on the backend complements app analytics.
Turning metrics into action
Pick one primary metric for the current phase: activation during early growth, retention once you have users, revenue once retention is healthy. Review it weekly with a small set of supporting metrics, investigate changes, and form hypotheses you can test. Our mobile app development projects include an event plan as part of the specification, so measurement is in place on launch day.
Key takeaways
- Ignore vanity metrics like total downloads; track metrics that point to decisions.
- Define an activation event and measure how many new users reach it.
- Read retention as cohort curves matched to your app's natural usage frequency.
- Monitor stability closely and plan events and privacy before you build.