Nuxie

Free app growth tools

Paywall A/B test calculator

Find out how many users you need and how long a paywall test could take before you build another variant.

Two variants · 50/50 allocation · 95% confidence · No signup required

Plan your experiment

Unique exposed users who convert within your chosen window.

20% lift takes a 5% baseline to 6%, a 1 percentage point increase.

Total across both variants. Count each user once, not every paywall view.

Time each user has to convert. Include trial time if measuring paid conversion.

Your planning estimate

8,158

unique users per variant · 16,316 total

Target conversion
6% from 5%
Estimated time to a mature result
28 days

21 collection days + 7 days for the final users to convert. Collection rounds up to complete weeks, with a two-week minimum.

How to use this estimate

Choose one binary outcome, such as paid subscription within 14 days of first seeing the paywall. Measure your baseline using that same outcome, audience and window. Assign each eligible user to one variant and keep their assignment stable.

The minimum detectable lift is a planning assumption, not a predicted result. At 80% power, a test has approximately an 80% chance of detecting an effect of that size under these assumptions. 95% confidence corresponds to a 5% significance threshold; it does not mean a winning variant has a 95% chance of being better.

Decide the sample target and stopping rule before starting. Wait until both variants reach the target and all conversion windows close. Repeatedly checking significance and stopping early invalidates this fixed-horizon calculation. Sequential tests need a different method.

Method and limits

We use the two-sided normal approximation for two independent proportions, with equal group sizes and no continuity correction. The method follows UBC’s sample-size calculator. Sample size per variant is rounded up.

n = [1.96√(2p̄(1−p̄)) + z(power)√(p₁(1−p₁)+p₂(1−p₂))]² / (p₂−p₁)²

Here p₁ is baseline conversion, p₂ is the target, and p̄ is their average. The two-week minimum is a practical scheduling assumption to cover weekly patterns, not a statistical guarantee. Traffic changes, unequal allocation, repeat users, multiple variants, multiple outcomes and very rare conversions need additional planning. This estimates conversion differences, not revenue per user or long-term retention.

Build the experience behind the test

Nuxie helps you create and deliver in-app Experiences. Explore Experiences or see pricing when you’re ready to build.