An install is not a user. It is a person who tapped a button in a store listing and is now deciding, in about ninety seconds, whether your app deserves a place on their home screen. Most onboarding advice ignores that pressure and treats the first launch like a guided tour of features nobody asked for.
We install a lot of apps on mid-range phones with old batteries and patchy signal, because that is the device most people actually carry. The apps that survive that test share a pattern: they get to the first useful action quickly, ask for trust only after they have earned it, and make the empty state feel like a beginning rather than a failure.
Why the first session decides the rest
Retention charts usually show a steep drop in the first three days. The cause is often visible in the first three minutes. A user opens the app, meets a carousel, creates an account, grants four permissions, and lands on an empty dashboard with no idea what to do next. Nothing has gone wrong technically, but nothing has gone right either.
The first session has one job: produce a small, real result. A saved note, a scanned receipt, a completed workout, a tracked package. Once a user has one item of their own inside the app, the reason to return is concrete. Features can be discovered later; the first win cannot wait.
Measure time-to-first-value
Define the first value moment in plain language and measure the time it takes a new user to reach it. For a budgeting app it might be the first categorised transaction. For a language app it might be the first completed lesson. If the median is over two minutes, onboarding is doing too much before it does anything useful.
Pair that with a completion rate for the first session and a day-seven return rate. Installs and daily opens are easy to inflate and hard to act on. Time-to-first-value tells you where the friction is, and a short screen recording of five real sessions usually shows the same two or three obstacles.
Permissions before trust
A permission prompt is a request for access to something personal. Ask for the camera before a user has tried the scanner and the answer is often no. Ask at the moment the feature is used, with one sentence explaining the benefit, and the opt-in rate changes without any dark pattern.
Notifications are the worst offender. An app that asks on first launch, before delivering any value, trains people to decline. Wait until the user has completed a meaningful action, then explain what the notification will contain. A reminder about a saved item is easy to accept; a vague promise of updates is not.
Trust is not a splash screen. It is what the app does before it asks for anything.
Sign-up, guest mode and account walls
An account wall in front of the first screen is a tax on curiosity. Unless the product genuinely cannot work without an identity, offer a guest mode and let people create an account when they want to sync, share or recover their work. The account becomes a feature, not a toll booth.
When sign-up is required, reduce it to the smallest useful form. Email and password is fine; a password rule with six unexplained constraints is not. Social login helps only if the provider is one your audience actually uses, and it must not silently request more data than the app needs.
Empty states and the first-run checklist
An empty screen is a design problem, not a neutral default. Show one clear action, a sample of what the screen looks like when it is full, and a way to start without creating content from scratch. Templates and demo data are cheap and they remove the blank-page hesitation that kills first sessions.
A short checklist can work if it is honest about length. Three items, visible progress and the ability to skip is helpful. Ten steps with a progress bar that barely moves is a chore. The checklist should describe the user's goal, not the app's feature list.
- One primary action on the first screen, not four equal buttons.
- A sample or template so the empty state is not a dead end.
- Permissions requested at the moment of use, with a reason.
- Guest mode available before an account is required.
- A skip option on every tour that is not essential.
Patterns that fail quietly
The five-slide carousel is the most common. People swipe through it without reading, then arrive at the same empty screen. Tooltips that cover the interface are a close second: they explain a button by hiding it. Both patterns feel like onboarding to the team that built them and like an obstacle to the person holding the phone.
Another quiet failure is the reward before the work. A streak counter or a badge on day one means nothing because the user has not done anything yet. Save the celebration for the first real result, and make it specific to what the person accomplished.
A test plan for the next release
Pick one onboarding change and measure it against the current flow for two weeks. Remove a permission prompt, add guest mode, or replace the carousel with a single sample item. Keep the change only if time-to-first-value falls and day-seven retention holds or improves. One change at a time is slower to plan and much faster to learn from.
For the pricing side of the same product, read our guide to mobile app subscription models. For the infrastructure behind it, see our Cloud Computing section.
