Open the app. A full-screen "Create your account" form. Fill it in, then swipe through five intro cards. A location prompt, then a notifications prompt, both denied because nobody said why. Then a dashboard full of empty charts and the words "No data yet". That's about a minute, and it's where plenty of new users close the app for good.
We see this sequence in product reviews all the time, and it's almost always an onboarding UX problem rather than a product problem. The features are fine. People just never get far enough to find them.
It's also a cost problem. Each person who leaves in the first session took your acquisition spend with them: the ad click, the store listing work, the referral. Most of the fixes are small and testable, which makes this one of the cheaper retention problems to work on.
Users arrive with expectations you didn't set
Nielsen Norman Group's summary of Jakob's Law puts it plainly: people spend most of their time in digital products other than yours, and those products set their expectations. If your first screen behaves differently from the apps they already use, they feel the extra effort straight away.
The platform owners have also moved against heavy upfront onboarding. Apple's Human Interface Guidelines say to delay sign-in for as long as possible, because people often abandon apps that make them sign in before they can do anything useful. Android's guidance is to ask for a permission in context, when the user starts to interact with the feature that needs it.
NN/g's analysis of mobile-app onboarding (21 June 2020) found that deck-of-cards tutorials didn't improve task performance. Its advice is to avoid instructional onboarding wherever you can and put the effort into a more usable interface instead.
Five onboarding UX fixes that protect user retention
Each fix has a before and after. These are illustrative patterns from reviews, not client results, so treat each "after" as a hypothesis to test in your own product.
1. Show something useful before the sign-up form
Before: The app opens on "Create your account". Nothing is visible behind it.
After: The app opens on something people can browse, such as sample content, a demo project or the live catalogue. Sign-up appears when they try to save, buy or share.
Apple's own example is a shopping app that lets people browse as much as they want and asks for sign-in only at purchase. When you do need an account, Apple suggests Sign in with Apple or, failing that, passkeys, so nobody has to invent a password in their first session. One practical gotcha: guest state has to survive sign-up. If the cart or draft disappears when the account is created, you've just punished the user for trusting you.
2. Swap the intro carousel for one real task
Before: Five swipeable cards describe features the user hasn't needed yet. "Skip" sits small in the corner.
After: The user lands in the real interface with one guided task, such as "Create your first invoice", and a short tip next to the control that matters.
Apple's onboarding guidance says people grasp and retain information better when they actually perform the task than when they view instructions. If you keep a tutorial, make it optional, don't show it again once it's been skipped, and put it somewhere findable later, like help or settings.
3. Ask for permissions when the feature is used
Before: Notification, location and camera prompts fire back to back on first launch.
After: The camera prompt appears when the user taps "Scan receipt", after a one-line explanation of why. Notifications are requested once the user has done something worth being notified about.
Timing matters more than most teams realise. On Android, if a user taps Deny for the same permission more than once, the system treats it as a permanent denial and stops showing the dialog. A badly timed prompt can switch a feature off for that user for good. When testing, adb shell dumpsys package shows a permanently denied permission flagged as USER_FIXED, and Android's docs show how to clear the flag so testers see the dialog again. Android also asks apps to degrade gracefully after a denial rather than block the user.
4. Cut the sign-up form to what the first session needs
Before: The form asks for company size, role, phone number, industry and how they heard about you.
After: It asks for an email address, or uses a passkey. Profile questions move to where they're useful: industry when it changes a template, role when inviting a teammate.
This is progressive disclosure, which NN/g described back in 2006: show a few of the most important options first and keep the rest for when users ask. Marketing data you want from everyone can usually wait until the user has a reason to stay.
5. Make empty screens do some work
Before: After sign-up, the dashboard shows blank charts and "No data yet".
After: The empty dashboard says what will appear there, offers one primary button ("Connect your bank feed" or "Import a CSV"), and can show sample data the user clears with one click.
NN/g's guidance on empty states is to use them to communicate system status, provide learning cues and give a direct path to the task that fills them. A blank screen leaves people wondering whether the product is even working.
A seven-question onboarding audit
Run this in your next product review. Every "no" is a candidate experiment.
- Can a new user reach something useful before creating an account?
- Is there one clear first task they can finish in the first session?
- Is every permission requested when its feature is used, with a reason?
- Does the sign-up form ask only for what the first session needs?
- Does every empty screen explain itself and offer a next action?
- Can users skip any tutorial and find it again later?
- Are you measuring the first session step by step, not just sign-ups?
Number 7 is where most teams come unstuck. You can't fix app drop-off you can't see. Fire an analytics event at each onboarding step with a consistent naming scheme (for example onboarding_step_viewed and onboarding_step_completed with a step property), then compare cohorts before and after each change.
| Signal | What it tells you | Where to look |
|---|---|---|
| Step-by-step funnel completion | Which screen loses the most people | Product analytics events |
| Time to first key action | How long it takes to reach value | Event timestamps per user |
| Permission grant rate by prompt | Whether timing and wording work | App analytics, platform consoles |
| Day 1 and day 7 return rate | Whether the first session built a habit | Cohort retention reports |
| Support tickets from new users | Where the interface fails to explain itself | Help desk tags |
Our onboarding review, week by week
We don't redesign anything at first. We pull your funnel data, watch a handful of first-time users go through the flow (recorded sessions or moderated tests), and walk every screen against the Apple and Android guidance above. You get a ranked list of friction points, each with a rough effort estimate and the metric we'd expect it to move.
After that, our UX and product design team prototypes the top two or three changes, tests them with real users, and works with your engineers to ship them behind feature flags so you can measure the difference. Our designers and developers sit in the same team across Melbourne and Sri Lanka, so there's no long handover from prototype to production.
The temptation is to rebuild the whole flow at once. Resist it. Change one step, measure, then move to the next. If screens look different from one flow to the next, a small design system helps. If a new user can't finish sign-up with a keyboard or screen reader, read our WCAG 2.2 basics next.
Sign-ups healthy but active users not following? Look at the first minute. Book a UX audit and we'll show you where new users stall and what to change first.
Frequently asked questions
What is onboarding UX?
Onboarding UX is the design of a new user's first experience with a product, from opening the app or site to finishing the first meaningful task. It covers sign-up, permissions, first-run guidance and empty states. Good onboarding gets people to the point where the product proves its value as fast as possible, with as little explanation as it can manage.
Should my app have an onboarding tutorial?
Usually not. Nielsen Norman Group recommends avoiding instructional onboarding where possible and improving the interface instead, because swipe-through tutorials didn't improve task performance in its research. A tutorial can help when a feature works unlike common patterns or is genuinely new. If you include one, keep it short and skippable, and let people find it again later.
When should an app ask for permissions?
Ask when the user starts to use the feature that needs the permission, and explain the benefit first. On Android, denying the same permission more than once counts as permanent, so a badly timed request at first launch can disable a feature for that user. Apple suggests putting a request in onboarding only when the app can't function without it.
How do I measure whether onboarding changes improve user retention?
Track each onboarding step as an analytics event, then compare cohorts before and after each change. Useful measures are step completion, time to the first key action, permission grant rates, and day 1 and day 7 return rates. Change one thing at a time where you can, so you know which fix moved the numbers.
UX & Product Design
Book a UX audit
Tell us where you're stuck and we'll come back with practical next steps.
Get in touchSources
- Nielsen Norman Group: 10 Usability Heuristics for User Interface Design
- Nielsen Norman Group: Mobile-App Onboarding: An Analysis of Components and Techniques
- Nielsen Norman Group: Progressive Disclosure
- Nielsen Norman Group: Designing Empty States in Complex Applications: 3 Guidelines
- Apple: Human Interface Guidelines, Managing accounts
- Apple: Human Interface Guidelines, Onboarding
- Android Developers: Request runtime permissions