Your retention problem is a first-session problem.

The standard retention playbook — push campaigns, streaks, win-back emails — operates on users who have already decided to leave. Open your analytics and look at where the curve actually falls: for most apps the biggest single drop is between install and the end of the first session. That's not a retention problem you fix in week three. It's a first-five-minutes problem, and it's fixable this week.

The cliff is earlier than you think

Retention gets framed as a long game — keep people engaged, build habits, send the right notification on day twelve. But pull up your own day-by-day numbers and the story is almost always the same shape: a cliff at day one, then a slope that flattens. The users who come back on day two mostly keep coming back; the ones who don't were usually gone before the first session ended.

That shape tells you where the leverage is. Improving day-30 retention by a point is a grind against users who barely remember installing you. Moving the day-1 number means intervening at the only moment you're guaranteed to have: the one session everyone gives you. A user in their first session has no investment in your app, no data inside it, no habit around it, and a switching cost of exactly zero. They arrived with one question — does this thing do what the store page said? — and they will answer it fast, with whatever you put in front of them.

Re-engagement can't fix a broken first impression

The uncomfortable part: most of the retention tooling you're tempted to reach for assumes a relationship that doesn't exist yet. A win-back notification to someone who never reached your app's value isn't a reminder — it's an ad for a product they already tried and rejected. You're re-marketing to a "no."

Mechanically, you often can't even reach them. On iOS you get one shot at the system notification prompt — decline it and you can't trigger it again; the user has to dig into Settings themselves. Android has required the same explicit runtime permission since Android 13. Apps that fire the permission dialog on first launch, before showing anything worth being notified about, collect a denial and quietly lose the channel they were planning to win users back with. The re-engagement strategy dies in the same broken first session it was supposed to rescue.

Find your activation moment

To fix the first session you need to know what it's for. Every app has an activation moment: the earliest action after which a new user actually comes back at a meaningfully higher rate. For a screenshot tool it's exporting the first finished design. For a habit tracker it's completing a check-in, not creating an account. For a photo editor it's saving one edited photo they like. It is almost never "finished the tutorial" or "granted all permissions" — those are your goals, not the user's.

If you have analytics, finding it is a small job: take a recent cohort of new users, split them by what they did in session one, and compare week-one return rates across those groups. The action whose presence most separates returners from ghosts is your activation moment. If you don't have analytics, logic gets you most of the way: it's the moment the user first receives the value your store listing promised. Not configures it, not previews it — receives it.

Rebuild the first five minutes around it

Once you know the target, the redesign is a counting exercise. Cold-start the app as a new user and count every tap, screen, and decision between launch and the activation moment. Each one is a toll booth, and some fraction of users turns around at every single one. Then start removing:

Move the sign-up wall behind the value. Unless the product literally cannot function anonymously, let the user do the thing first and create the account when there's something worth saving. "Sign up to see if this was worth signing up for" is the most common toll booth and the most expensive one.

Ask for permissions in context, or not at all. A notification prompt makes sense the moment there's something concrete to be notified about — a finished export, a reminder the user just scheduled. On launch it's noise, and on iOS it's a one-shot you're wasting on someone who met your app forty seconds ago.

Kill the tutorial carousel. Five swipeable cards explaining the interface is a confession that the interface doesn't explain itself. Users swipe through without reading — you've added friction and taught nothing. Teach in place, at the moment a feature becomes relevant.

Make the empty state do the onboarding. The first screen most new users actually study is your emptiest one. A blank canvas with a lonely plus button outsources the hardest question — "what do I do now?" — to the person least equipped to answer it. Sample content, a pre-populated example, or a single unmissable next action turns the empty state from a dead end into a running start. The user's first act should be editing something, not staring at nothing.

Your listing is screen zero of onboarding

Here's the part that gets missed because it lives in a different tool: the first session doesn't start at launch. It starts on the store page. Your screenshots make a promise, and the first session either keeps it or breaks it. When a listing sells outcome number one but the app opens on a configuration wizard, users don't experience "setup" — they experience a bait and switch, and they're gone before your carefully deferred sign-up wall even renders.

Two practical consequences. First, your screenshot order should roughly mirror time-to-value: the outcome in frame one should be the outcome a new user can reach fastest, because that's the promise they'll test. Second, when you shorten the path to value in the app, update the listing to lead with it — the promise and the first session are one funnel, and they need to be designed together, by the same person, with the same story.

What to watch after you ship it

Three numbers tell you if the rebuild worked. Time to first value — median minutes from first launch to the activation moment; it should drop sharply. Step-level drop-off — what fraction of new users clears each remaining screen between launch and activation; any step shedding a big slice is your next deletion candidate. And day-1 retention split by activation — activated versus not. If activated users return and non-activated ones don't, you've confirmed the moment is real, and the whole game becomes getting more people to it, faster.

One session. That's the honest budget every new user gives you. Spend it on their first win, not your form fields.