Six onboarding flows tested: 0.8% to 1.29% conversion (+61%)
Commentary byLiftArchiveOur analysis, not theirsThe best of six onboarding flows lifted conversion to subscription from 0.8% to 1.29% (+61%) and ARPU from $0.12 to $0.14 (+16%).
AEZAKMI Group ran the test on its LED Light Controller app, and the case study was published on the Apphud blog.
The text does not describe what the winning onboarding contained.
- Conversion to subscription: 0.8% to 1.29%, a 61% relative lift.
- ARPU: $0.12 to $0.14, a 16% lift. ARPU grew less than conversion.
- Same flow, different phases: one option converted 1.12% in phase one and 1.29% in phase two with nothing changed, a reminder that results drift over time.
Control
The app's original onboarding flow.
Variant
Six onboarding options in total. Because the tool ran at most five variants at once, the test was split into two phases, and the best flow went live.
Why it likely won
Our reading of the result, not the practitioner's claim.
- Onboarding sets up the first paywall view, so a flow that shows the app's value faster can lift conversion. We can't say which change did it, since the screens are not described.
- The smaller ARPU gain suggests the extra subscribers were worth a bit less each, perhaps through cheaper plans or more early cancellations. This is our guess; the source does not explain it.
- Testing many flows side by side raises the odds of finding a better one, but also the odds that one wins by chance.
Caveats: Published on the blog of Apphud, which sells the experiments tool used, so it may lean positive. Sample size and duration were not shared, and the content of the winning flow was not described.
Who ran it
AEZAKMI Group ran this test. Source: Apphud blog. The numbers are theirs; the summary and commentary on this page are ours.
How to run this test on your app
- Hypothesis
- If we test several onboarding flows at the same time against our current one, then at least one will lift conversion to subscription, because onboarding shapes how users see the app's value before the paywall.
- Primary metric
- ARPU per new user
- Guardrails
- Onboarding completion rate
- Trial-to-paid conversion
- Day-7 retention
- Who should test it
- Apps with enough installs to split traffic many ways and still reach significance. Smaller apps should test two or three flows at a time, or the test will take too long.
Questions
Why did the same flow score differently in each phase?
Traffic quality and season change over time, even when the flow does not.
That is why options should run at the same time, and why winners from different phases should be compared with care.
How many onboarding variants should I test at once?
Only as many as your traffic supports. Each extra variant splits users further and makes a false winner more likely.