Designing Onboarding Flows That Convert Without Overwhelming Users
On a recent fintech app, we inherited an onboarding flow with nine screens before a user reached the home dashboard: welcome, three permission requests, a tutorial carousel, account type selection, and two form screens. Drop-off between screen one and the dashboard was 61%. We instrumented Amplitude events on every screen transition and found the biggest single-screen loss (22%) was the notifications permission prompt, asked before the user had done anything to justify wanting notifications.
The fix was reordering, not removing. We moved permission requests to the moment they're actually needed — camera access when the user taps 'scan a receipt,' not on screen two. We also collapsed the three-screen tutorial into contextual tooltips shown inline during first use. Total screens before dashboard dropped from nine to three, and completion rate rose from 39% to 68% over an eight-week rollout.
Deferred account setup is the other lever we reach for. Letting users explore a read-only version of the product before requiring signup increased signups themselves in one case, because people had already seen enough value to commit. We gated write actions — saving, purchasing, syncing — behind auth, not browsing.
Not every flow should be minimized blindly. For a healthcare app with regulatory consent requirements, we couldn't defer disclosures, so we focused effort on copy clarity and progress indicators instead — a five-step flow with a visible 'step 3 of 5' counter converted better than the same flow with no indicator, even though both had identical content.
The measurement discipline matters more than any specific pattern. We ship onboarding changes as A/B tests where the platform allows it, and track completion by cohort, not just an aggregate rate, because power users and first-time downloads behave very differently on the same screens.
