In this guide
Most apps lose the majority of new users early. You cannot fix that with a bigger number on a dashboard. You fix it by knowing what a "successful first session" is for your app, finding where people stop, and changing one thing at a time. This guide gives you that method. It contains no benchmarks, because your category, audience and traffic source decide what is normal.
Step 1: Define activation
Activation is the first moment a user gets the value you promised. Write it as one observable event.
- Notes app: created and saved a first note.
- Fitness app: finished a first workout.
- Budget app: added a first transaction.
If you cannot name the event, the listing, the onboarding and the app probably promise different things. Fix that first.
Step 2: Read day 1, day 3 and day 7
Day 1 retention is the share of a install-day cohort that opens the app the next day. Day 7 works the same way. Look at the shape, not a single value:
- Big drop before day 1: a problem in the first session. Look at onboarding, permissions, sign-up and crashes.
- Fine day 1, falling by day 3-7: the app was understood but gave no reason to return. Look at habit, content, reminders.
- Different cohorts, different curves: compare by traffic source, version and country before you draw conclusions. Small cohorts are noisy; do not react to a few users.
Step 3: Find the drop-off point
Build a simple funnel from install to activation: first open, onboarding finished, permission answered, activation event. The step with the largest loss is your candidate. Pair the numbers with a few real sessions: watch someone use the app, or ask testers where they hesitated. A funnel says where, people say why. See onboarding questions for testers.
Step 4: One change per week
- Write the hypothesis: "Users stop at step X because Y."
- Make one change that addresses Y.
- Compare the next cohort with the previous one, same traffic source if possible.
- Keep, revert, or try the next idea.
Several changes at once make the result unreadable. With small user numbers, treat results as direction, not proof.
Worksheet
| Question | Your answer |
|---|---|
| Activation event | |
| Where is the largest loss? | |
| Hypothesis (why?) | |
| This week's one change | |
| How will I judge it, and when? |
Common traps
- Counting installs from sources that do not match your audience, which makes retention look worse than the app deserves.
- Changing the activation definition after seeing the numbers.
- Asking for notification permission before any value is shown. See when to ask for push permission.
Private, structured feedback from testers before launch shows many first-session problems early. On AppReviewX, one feedback you give earns one credit, and new accounts start with 0 credits.
Related: ASO and launch hub, Mobile app onboarding, Push notification permission timing.