App Onboarding: Best Practices Backed by A/B Tests

App onboarding is the set of screens a new user sees between first launch and real use of the app.

For a subscription app, its job is to get the user to one meaningful action and to a trial or purchase decision in the first session.

In our test database, the onboarding changes that won asked about the user's goals, showed the product's value before the price, and put the paywall inside the flow rather than after it.

This guide follows a new user's path in order: what onboarding has to achieve, the main flow types, how long it should be, where permission prompts go, where the paywall goes, what top apps have tested, and how to test your own flow.

Every example links to a real result in our app onboarding test library.

What onboarding must achieve: activation and the trial start

Onboarding has two jobs, and they have to happen close together.

The first is activation: the user does the one thing that shows them the app works for them.

The second is the money decision: a trial start, a purchase, or a clear move to the free tier.

The timing matters more than most teams expect.

RevenueCat's State of Subscription Apps 2025 reports that 82% of trial starts happen on the same day the user installs the app.

So the first session is where most trials are won or lost, and onboarding is most of the first session.

That means you should judge onboarding on more than completion rate.

A flow that every user finishes but nobody pays after has failed. Use three numbers together:

• Install to trial (or install to paid): the main conversion metric for the flow.
• Trial to paid: catches flows that push weak trials that later cancel.
• Early retention or activation: did the user reach the core action, and did they come back?

The ASL Bloom redesign by Applica shows all three moving at once.

Applica added questions about each user's learning goals and changed the paywall to two screens.

The result was +15% trial conversion, +12% ARPU and +18% 30-day retention.

RevenueCat's write-up also reports a 28% rise in users who completed their first real-world communication goal.

The changes shipped together, so we can't credit any one screen.

Onboarding flow types: walkthrough, quiz, value demo, sign-up first

Most onboarding flows are built from four patterns.

Many apps mix them, but each has a different purpose and a different risk.

Walkthroughs

The classic feature carousel is the cheapest flow to build and the easiest to ignore.

Apple's Human Interface Guidelines on onboarding suggest considering context-specific tips instead of a single upfront flow.

They also say that if you offer a separate tutorial, it should be optional and easy to find later.

A short carousel can still help when it sets up the paywall.

In one test shared by Rosie Hoggmascall, adding a three-slide carousel before the paywall lifted trial opt-in from 8% to 15%.

We cover that test in the paywall section below.

Quizzes

The quiz is the dominant pattern in health, fitness and self-improvement apps.

It asks about the user's goals, then reflects the answers back as a "plan" right before the paywall.

Our onboarding quiz element page collects every test of this pattern.

The best-documented quiz win is Five Minute Journal.

Applica rebuilt the onboarding around user research, framing the app as a fix for each user's stated problem and adding testimonials early.

The result was +20% ARPU, about +10% trial starts and about +14% trial conversions. The changes were bundled into one redesign.

Value demos

A value demo lets the user see the output before paying: a sample edit, a generated result or a preview of the program.

The evidence for this pattern in our database comes mostly from paywalls that act as demos.

Clear 30 rewrote its paywall to preview the program and saw conversion rise from 20% to 30% (+50%), a before-and-after comparison, not a split test.

Kovalee tested a 10-second video preview before the PepTalk paywall, and both video variants trended above control.

Sign-up first

Asking for an account before showing anything is the riskiest pattern for a subscription app.

It adds a step before the user has a reason to take it.

Apple's App Review Guidelines (5.1.1(v)) also say that if your app doesn't include significant account-based features, you should let people use it without a login.

A common compromise is to collect the account after the paywall or after the first saved result, when there is something worth protecting.

Our sign-up wall element page tracks tests of where the account step goes.

Onboarding length: how many screens?

The usual advice is "keep it short." The test data for subscription apps points the other way, with a caveat: longer works when each screen earns its place.

Lose It! kept adding questions to its onboarding and saw its trial start rate rise by double digits.

In RevenueCat's write-up, the team said they kept making it longer "until we got diminishing returns."

The same article describes Me+, a top health and fitness app in the UK, as having about 45-50 onboarding screens that take around 7-8 minutes.

Why would more screens help? Three reasons show up again and again in the tests we track:

• Investment: each answer is a small commitment, and users who have answered ten questions are more likely to see the plan they built.
• Personalization: the paywall can repeat the user's goal back to them, which makes the price feel tied to an outcome.
• Self-selection: users who finish a long flow are more motivated, so the people reaching the paywall convert better.

The caveat is that the third reason can hide a loss.

If a longer flow cuts the number of users who reach the paywall by more than it raises their conversion, total trials fall.

Always measure install to trial, not paywall conversion alone.

Apple's onboarding guidance pulls toward brevity: a "brief, enjoyable experience that doesn't require people to memorize a lot of information."

These two views don't really conflict. A 30-screen quiz where each screen is one tap is quick to finish.

A five-screen tutorial that asks the user to remember gestures is not.

A practical rule from the tests we have: there is no correct screen count.

Start with the questions you will actually use, add social proof and a results preview, and test longer and shorter versions against install-to-trial.

Permissions inside onboarding

Notification, tracking, location and camera prompts are where many onboarding flows leak users.

Once a user taps "Don't Allow," they have to go into Settings to change it, so timing matters.

Both platforms push you toward asking in context:

• Apple's HIG says that if the app needs access to private data before it can work, you can include the request in onboarding and explain the benefit. Otherwise, ask when people first use the feature that needs it.
• Apple's notification documentation gives the example of a task app asking after the user schedules a first task, rather than on first launch.
• On Android 13 and later, notifications are off by default for new installs. Android's notification permission guide recommends letting users get to know the app first, or asking when they tap something like an alert bell.

There are also hard rules. Under App Review Guideline 5.1.1(iv), apps must not "manipulate, trick, or force people to consent to unnecessary data access."

Guideline 5.1.2(i) says apps may not require users to enable push notifications, location or tracking to use the app or to get any reward.

Tracking itself needs explicit permission through the App Tracking Transparency prompt; see our ATT guide.

Within those rules, two techniques are widely used:

• Permission primers: a custom screen before the system dialog that explains what the user gets, with a "Not now" option that skips the system prompt. Our permission primer element page tracks tests of this pattern.
• Provisional notifications on iOS: with provisional authorization, the app can send notifications quietly to Notification Center without a prompt, and the user chooses to keep or turn them off when they see one.

We don't yet have a split test in our database that isolates permission timing for a subscription app, so treat this section as platform guidance, not a measured result.

For what to send once you have permission, see our push notifications guide.

Where the paywall goes in onboarding

Paywall placement is the most heavily tested onboarding question in our database, and the results are consistent: showing the paywall inside onboarding, to everyone, beats waiting for users to find it.

The open question is how much value to show first.

Our paywall timing element page collects all of these tests.

Paywall in onboarding vs. paywall later

Greg moved from usage-based limits to feature gating and started showing the paywall to every user in onboarding.

The result was about +400% trial sign-ups, with sign-up-to-trial going from 3% to 15%.

Rootd moved its paywall to the start of onboarding and reports revenue up 5x.

That was a before-and-after comparison, so other changes in the same period may have helped.

FitnessAI, working with Jake Mor, put the paywall before onboarding with a video and saw +50% more users seeing the paywall.

That is the plain mechanism behind most of these wins: a paywall placed after onboarding is only seen by users who finish it.

Paywall before onboarding, but after a little context

The cleanest placement test we have was shared by Rosie Hoggmascall on RevenueCat's blog. Three flows were compared:

1. Welcome, onboarding, home, then paywall: 2% trial opt-in.
2. Welcome, paywall, onboarding, home: 8% trial opt-in.
3. Welcome, a new three-slide carousel, paywall, onboarding, home: 15% trial opt-in.

See the full test. Moving the paywall earlier quadrupled opt-in, and adding a short value carousel before it almost doubled it again.

Two paywalls in one flow

Rizz App tested two paywalls in onboarding: one right after the landing screen with the trial toggle off by default, and a second at the end of the flow.

Superwall's case study reports +10% onboarding conversion and +10% trial conversion. Results are vendor-reported.

Where hard paywalls go wrong

Earlier and harder is not always better. BoldVoice tested a hard paywall in onboarding with no close button.

It got more trial starts, but refunds rose by more than the gain.

Wizz went the other way, showing fewer offers at better moments, and saw +81% on the primary offer flow with LTV up about 20%.

The lesson is to judge placement tests on revenue after refunds and renewals, not trial starts.

For the wider choice between gated and open models, see hard vs. soft paywalls, and for the screen itself, our paywall guide.

One rule applies wherever the paywall sits. App Review Guideline 3.1.2(c) says that before asking someone to subscribe, you should clearly describe what they will get for the price.

Onboarding examples from top apps

Here are the onboarding tests in our database with real numbers, and what each one changed.

A few patterns stand out across these tests:

• Personalization wins, when it's real. Both Applica redesigns asked users what they wanted and then shaped the rest of the flow around the answer.
• Benefits beat features. Five Minute Journal moved from listing features to describing the result for the user.
• Testing many flows pays off. AEZAKMI Group's LED controller app tested six options and lifted conversion to subscription from 0.8% to 1.29%. Apphud's write-up also notes that one unchanged control converted at 1.12% in the first round and 1.29% in the second, which shows why variants should run at the same time, not one after another.
• Onboarding doesn't have to live in the app. Teaching.com ran onboarding on the web before download and saw about +50% conversion to trial. The team credited its well-known brand, which made users comfortable paying on the web. Our web-to-app guide covers this funnel.

Two of the strongest results here are bundled redesigns, so they tell you a direction, not which screen did the work.

If you copy a pattern, test it against your current flow rather than assuming the same lift.

How to test your onboarding flow

Onboarding is one of the easiest parts of an app to test, because every new user passes through it.

Here is a simple process that fits the tests above.

1. Map the flow and measure each step. Track the share of installs reaching each screen, the paywall view rate, and install to trial. The biggest drop is usually your first test.
2. Pick one primary metric. Use install to trial or install to paid. Watch trial-to-paid, refunds and day-7 or day-30 retention as guardrails, because BoldVoice's result shows trial starts alone can mislead.
3. Test big changes first. Paywall placement, quiz vs. no quiz, and web vs. in-app onboarding have produced the largest moves in our database. Button copy rarely does.
4. Run variants at the same time. Sequential before-and-after comparisons pick up seasonality and ad-mix changes, as the AEZAKMI control showed.
5. Size the test before you start. Small apps often need weeks to detect a 10% lift. Use our sample size calculator to check what's realistic.
6. Wait for trials to convert. A placement test can look like a win at day 3 and a loss at day 30. Read the result after trials end.

Tools differ in how they run onboarding tests.

Some paywall platforms can show different paywalls at different onboarding steps without an app update, while a fully new onboarding flow usually needs your own feature flags.

Our app A/B testing guide covers setup and pitfalls, and the app onboarding checklist turns this page into steps you can work through.

To compare your results with other apps, see our app trial conversion rate benchmarks.