App Zero: Get Mobile Right the First Time is a series for PMs at established companies who've been handed a mobile mandate and don't want to blow it. This is the fourth and final post, and here’s the first.
TL;DR
Most apps lose the majority of new users in the first 24 hours. The product isn't the problem. The onboarding is. You have two different user types hitting the same install screen (existing account holders and cold installs), and a single flow that serves neither well.
Fix the permissions sequence, get users to their Aha moment before you ask for anything else, define an activation event that actually predicts retention, and stop asking personalization questions you can't act on immediately. Then treat onboarding like a product surface with its own roadmap and keep iterating.
The average mobile app loses more than 75% of new users within 24 hours1. Not because the core product is bad. Because the team that built it forgot one thing: few download an app in a calm, focused, ready-to-learn state of mind.
They download it while waiting for coffee or between calls. They get the app at 11 PM because something in their feed reminded them it existed. But after a week, more than 88% of them are gone, and after a month, it’s 94%2.
That window is tiny. And most apps spend it explaining themselves.
The Motivation Illusion
The install spike is not commitment. It’s curiosity. And curiosity runs out fast.
A new user has enough patience to get through maybe three friction points before they close the app and never come back. Most teams respond to this by building a six-screen walkthrough that asks for push notification access on screen two and introduces pitches features the user can’t access now.
This is the motivation illusion. You see high install numbers and assume those people are pulling for you. They’re not. They’re skeptical, distracted, and two back-taps away from something else.
The reframe is this: onboarding is not a tutorial. It’s a conversion funnel with no second chances. Every screen you add costs the user motivation they aren’t getting back.
Two Different People Just Downloaded Your App
Most growth teams think of “new users” as a single population. They’re not.
For any app with an existing web presence, two completely different people hit the same install screen at the same time.
The first one already has an account. They’ve been on your desktop product for months. They downloaded the app because they want their data in their pocket. They do not need a tour. They need to log in and see their stuff, and every screen you show them before that moment is something they resent.
The second one found you through the App Store or an ad. They have zero context, no account, and no reason yet to trust you. They need proof of value before they’ll hand over an email address, let alone payment info.
When you build one flow for both of these users, one of them churns at twice the rate. Usually the web user, because they’re the most irritated by being treated like a stranger. Fixing this requires actual intention at the top of the funnel:
Use deferred deep links in web sessions. If an existing web user taps “Get the app,” that link should carry a signal. Land them directly on a login screen and skip everything else. This is a growth cheat code.
For cold installs, let users experience something before you ask for an account. Duolingo starts a lesson before you’re registered. That’s not a growth hack. It’s a choice to earn the relationship before demanding it.
The Four Decisions That Actually Determine Whether Onboarding Works
Forget the six-screen walkthrough debate. Onboarding comes down to four specific decisions. Each one creates a leak if you get it wrong.
Permissions Timing
Asking for push notification access on screen one is an amateur mistake that’s all too common. It keeps happening because permissions feel like a logistics problem. They’re a trust problem.
When a user sees “Allow Notifications?” before they’ve done anything in the app, they have no idea whether your notifications will be useful or annoying. So they tap “Don’t Allow,” and you’ve lost that channel permanently. Getting it back requires a manual trip to Settings that almost nobody makes.
The rule: earn the ask. Stack-rank your permissions by how much value the user needs to have experienced before each one makes sense. Push notifications come after the first completed action, minimum. Camera or location access gets triggered at the exact moment the feature that needs it becomes relevant, with one sentence explaining why.
Getting to the Aha Moment First
Every app has a moment when a new user first understands why the product will matter to them. For a grocery app, it might be when a previous order reappears, saving them 20 minutes of list-building. For a password manager, it’s the first time autofill works without friction.
Most onboarding flows bury that moment under setup steps and feature previews.
Map the shortest possible path from install to that specific moment. Then cut every screen that isn’t on that path. Advanced settings, social features, tutorials for things the user hasn’t tried yet, all of it can wait. The job is to get them to the point where they think, "Okay, this is actually useful," before you ask for anything else.
Account Creation
Every field in your sign-up form costs you users who decided the trade wasn’t worth it. Email plus password as the only option in 2026 is a slow bleed.
Sign in with Apple or Google removes most of the friction and gives you verified contact info as a bonus. Passkeys are catching up fast for apps where security is part of the value.
Delayed sign-up (let users experience something before requiring an account) works well for discovery apps. It falls apart for apps where personalization is the core product, because you can’t personalize anything for a ghost. Know which one you’re building before you pick a strategy.
Define What “Activated” Actually Means
“User opened app twice in the first week” is not an activation event. It is noise.
A real activation event is specific, observable, and actually predicts whether someone will still be using your app in 30 days. For a grocery app, it might be “completed checkout with at least five items.” For a ride-share, it might be “booked a second time within seven days of the first.”
To find yours, pull retention data on your day-30 users and work backwards. What did they do in the first session that churned users did not? That action is your activation event. Build the entire onboarding flow around getting new users there as fast as possible. Everything else is decoration.
The Personalization Trap
Asking users what they want during onboarding can reduce early churn. It can also make them feel like they wandered into an HR intake form when they just wanted to try the app.
One rule covers most of this:
Only ask a question if the answer visibly changes what the user sees on the next screen. Not eventually. Immediately.
A fitness app that asks “What’s your main goal?” and routes users to different workout tracks based on the answer? That works. A budgeting app that asks five questions about financial habits before showing any interface, then presents the same generic dashboard regardless of what you said? That destroys trust faster than no personalization at all.
Questions should be visual and fast. A card selection with big tap targets takes ten seconds and feels interactive. A form with text fields and dropdowns takes ninety seconds and feels like homework.
Two questions maximum. And only if you can point to exactly what changes on screen for each possible answer. If you can’t, cut it.
Keep Working on Onboarding After Launch
Onboarding is the highest-leverage surface in a mobile app. A 10% improvement in activation rate compounds across every new user from that day forward. Most teams touch it once at launch and don’t go back.
This is where most of the real opportunity is sitting, unworked.
The sprint structure that actually moves the number:
Week 1: Find where users are actually leaving. If you don’t have event-level tracking on every onboarding screen, that’s step zero. You need to know which specific screen has the highest drop rate, not just that churn happens in the first 24 hours. Set up a funnel in Amplitude, Mixpanel, or PostHog that shows you screen-by-screen falloff. Find the worst one. That’s your first test.
Week 2: Write a real hypothesis. Not “we think this screen is bad.” Something you can actually test: “We believe users drop at the permissions screen because we’re asking for notifications before showing value. If we move that prompt to after the first completed action, we expect drop-off here to fall by 20%.” One variable. Resist the urge to change three things at once because you’re impatient to see results.
Week 3: Read the results and move. 95% confidence before calling it. If you don’t have enough traffic to hit significance in a week, either run it longer or accept that acquisition is the bigger constraint right now. That’s a real choice, not a failure. Apply the winner, find the next worst screen, repeat.
One other tool worth using: session recordings. Pull 20 sessions from users who churned on day one and watch them. The patterns are usually obvious. Someone rage-tapping the same button repeatedly, a permission prompt appearing before any value, a flow that assumes the user knows what a feature does before they’ve used it. You can see all of this in an afternoon, and it will tell you more than any A/B test result.
The onboarding flow you shipped at launch is probably the one your users are still running through today. Most teams never go back.
That’s the whole game.
App Zero is a 4-part series for product managers at established companies launching their first consumer mobile app.
“App Uninstall Report-2025 Edition.” AppsFlyer: https://www.appsflyer.com/resources/reports/app-uninstall-benchmarks-report
“User retention: How to successfully keep users and grow your app.” Adjust: https://www.adjust.com/resources/guides/user-retention