Onboarding A/B test: decide on trial starts or paid conversion?
Trial starts answer in weeks; paid conversion is what you care about but waits for the trial to end and needs more traffic. Decide on trial starts when the change is about getting people to the offer, and on paid when it could change who pays.
What an onboarding test can catch
On trial starts, when 5.8% of new users start a trial:
| Users a week | 2 wk | 4 wk | 6 wk | 8 wk |
|---|---|---|---|---|
| 5,000 | +22.6% | +16.0% | +13.0% | +11.3% |
| 10,000 | +16.0% | +11.3% | +9.2% | +8.0% |
| 20,000 | +11.3% | +8.0% | +6.5% | +5.6% |
| 50,000 | +7.1% | +5.0% | +4.1% | +3.6% |
| 100,000 | +5.0% | +3.6% | +2.9% | +2.5% |
On paid conversion, estimated from the same trial rate (install to paid taken as 2.0% for 5.8% trial starts, the RevenueCat State of Subscription Apps 2026 medians):
| Users a week | 2 wk | 4 wk | 6 wk | 8 wk |
|---|---|---|---|---|
| 5,000 | +39.2% | +27.7% | +22.6% | +19.6% |
| 10,000 | +27.7% | +19.6% | +16.0% | +13.9% |
| 20,000 | +19.6% | +13.9% | +11.3% | +9.8% |
| 50,000 | +12.4% | +8.8% | +7.2% | +6.2% |
| 100,000 | +8.8% | +6.2% | +5.1% | +4.4% |
Paid needs about three times as many people for the same change, and the decision waits for the trial to end plus a week for refunds.
Which to choose
- Trial starts: copy, questions, screen order, a value screen before the paywall, sign-up moved later. Watch paid conversion as a guardrail: more trials of people who never pay isn't a win.
- Paid conversion: anything that could change who starts a trial or what they buy: the paywall offer, the trial itself, a plan built from answers that leads to a different plan.
- Cutting screens: published tests where onboarding was cut down lost about 13% of conversion; condensing long reading screens into a short carousel won once. Cut steps whose answers are never used, not the ones that set up the value.