FREE EXPERIMENTATION TOOL

A/B test sample size calculator.

Plan for the effect that would matter. Estimate visitors per variation and time to reach your sample.

YOUR TEST PLAN—

visitors per variation

Total sample—

Traffic-based estimate—

Target conversion rate—

Reaching the sample does not guarantee a winner. Allow complete business cycles and conversion lag; use a fixed stopping rule agreed before launch.

What the calculation assumes.

A fixed-horizon frequentist comparison of two independent proportions, equal allocation, and a binary conversion event. It uses a normal approximation without continuity correction. It is not a sequential-testing or multi-variant calculator.

A 20% relative lift on a 2% baseline means 2.4% conversion, a change of 0.4 percentage points. Power is the chance of detecting the specified effect if it exists under the model.

Formula and method

n per arm = [zα × √(2p̄(1−p̄)) + zβ × √(p₁(1−p₁)+p₂(1−p₂))]² / (p₂−p₁)², rounded up. p̄ = (p₁+p₂)/2. zα uses α/2 for the two-sided test.

Method reference: UBC: comparing two independent proportions ↗.