App metrics that matter: retention, stickiness and churn for founders
Downloads say the least. DAU/MAU, cohort retention and churn defined cleanly, plus how to instrument events with Firebase Analytics.
Downloads are the number everyone quotes and the number that says the least. An app with a hundred thousand downloads and five hundred weekly users has a marketing story, not a business. The metrics that actually describe an app's health are activity, retention and churn - and they are cheaper to instrument early than to retrofit later.
This guide defines the core metrics and how to wire them up, using Google's Analytics for Firebase documentation, checked on 19 November 2026.
The core definitions, without the fuzz
- DAU / MAU. Daily (or monthly) active users: people who did something meaningful in the window. "Active" must be defined by you - opening the app is the loosest version; completing your app's core action is the honest one.
- Stickiness. DAU divided by MAU, as a percentage. It answers "of the people who use this in a month, how many use it daily?" - the habit signal.
- Retention. Of the users acquired in a given period (a cohort), what fraction is still active after N days? Day-1, day-7 and day-30 retention are the standard checkpoints. Retention is the metric that decides whether growth spending is investment or bonfire.
- Churn. The mirror: what fraction of users (or subscribers, or revenue) is lost per period. For subscription apps, churn compounds quietly; see subscription pricing models for the lifecycle states that drive it.
Instrumentation: events, not pageviews
App analytics are event-based. Firebase's documentation describes events as insight into what is happening in your app - user actions, system events, errors. Two practical points from the docs: Analytics logs some events automatically with no code, and there is no limit on the total volume of events your app logs. The design work is choosing which custom events represent your product's value: the purchase completed, the booking made, the document signed. Ten well-chosen events beat two hundred noisy ones, and event names are forever once data accumulates - pick a naming convention first.
Reading the numbers without fooling yourself
- Cohorts, not averages. Overall retention hides change. If last month's users retain worse than the month before, the total can still look fine while the product is getting worse.
- Match the metric to the model. A utility opened twice a year should not be judged on DAU. Define the usage frequency your product is built for, then measure against that.
- Pair every number with its denominator. "Retention is 40%" is meaningless without the cohort definition, the window and the active-user definition behind it.
- Revenue metrics need lifecycle states. For subscriptions, measure active, trial, grace-period and churned users separately; Apple's and Google's billing systems report these states and retention messaging can recover some before they leave.
Sources
- Google Analytics for Firebase documentation, log events: https://firebase.google.com/docs/analytics/events
Drafted with AI assistance from this source. Benchmark figures vary too much by category and source to quote responsibly here; establish your own baselines from your first cohorts.