Why Users Drop Off During Onboarding, and How to Find the Step

Users rarely leave onboarding because of the whole flow. They leave at one step: usually the one that asks for too much before the app has shown any value. To find it, look at where people stop in your funnel, watch a few people go through it, and read what support hears. Then check the cause before you change anything. The same drop-off can come from five different problems.

"Our onboarding has too many steps" is the most common explanation, and it is often wrong. Sometimes the number of steps is fine and one of them is the problem. Fixing the wrong thing costs a release and teaches you nothing.

How do you find the step where people leave?

1. Count at each step. In your analytics, list how many people reach each onboarding screen. The largest single drop is where to look, not the total.

2. Watch real sessions. If you have session recordings, watch a handful of people who left at that step. If not, ask three people who have never seen the app to go through it while you stay silent.

3. Read the support messages. Questions like "why does it ask for this?" or "it will not let me continue" point straight to the step.

4. Check the exact moment. Is it a permission prompt, a form field, a verification code, a payment screen? "Step 4" is not an answer. "The phone verification on step 4" is.

If you do not have step-level data, add it first. It is usually a small piece of work, and everything else depends on it.

What are the common reasons people stop?

1. The app asks for too much before it shows value

Account creation, permissions and profile questions come before the person has seen why they should care. The fix is usually sequence, not removal: show something useful first, ask later.

2. It is not clear why you are asking

A request for location, contacts or personal data with no reason attached feels risky. A short line explaining what it unlocks often changes the answer.

3. The number of steps is not visible

An unknown number of steps is harder to accept than a known one. A clear progress indicator removes the feeling of being trapped. In Ghost Social, we broke setup into micro-steps with clear progress indicators to minimise friction and prevent fatigue.

4. The method is not explained

If the app uses an idea people have not seen before, they get stuck on the first real action. In Break the Cycle, three short screens teach the core mechanic (capture the thought, reframe it, rate the emotion) before a user creates their first card.

5. There is no way forward when something fails

An import fails, a code never arrives, an email is rejected, and there is no alternative. If a step can fail, it needs a fallback. A manual entry option, a resend, a way to skip.

How do you confirm the cause before fixing it?

Write the cause as a testable statement: "People stop at the phone verification because they do not know why it is needed." Then check it in the cheapest way available:

• Ask three people to explain, in their own words, why the app is asking.

• Check whether the step fails for a share of users (a code that does not arrive is a bug, not a design problem).

• Compare drop-off on that step by platform, device or traffic source. A difference points at something technical.

Only after that decide whether the fix is a reorder, a reason, a fallback or a cut.

What to avoid

• Cutting steps at random. Some steps earn their place. The question is which one costs more than it gives.

• Redesigning the visuals first. A cleaner screen with the same demand will lose the same people.

• Changing several things at once. You will not know which worked.

What to do this week

1. Get step-by-step counts for your onboarding. If you cannot, add the events.

2. Mark the biggest single drop.

3. Watch three sessions or three people at that step.

4. Write the likely cause as one sentence.

5. Test that sentence with the cheapest check above.

FAQ

What is a good onboarding completion rate?

It depends on your product, audience and how onboarding is defined, so a single benchmark is not reliable. Compare your own steps over time, and compare the step with the biggest drop against the others.

Should I remove sign-up from onboarding?

Sometimes the better move is to let people see value first and ask for an account at the moment it matters. Whether this fits depends on what the account is for.

When should I get outside help?

When you can see the drop but cannot agree on the cause, or when the problem seems to be in several steps. A review of the full journey separates the causes and puts them in order.

If you want a second pair of eyes on your onboarding, see our UX audit for mobile apps, or book a free 30-minute Product Diagnosis. For the next question, what to ask and what to skip, read sign-up and onboarding: what to ask and what to skip.