All free tools
2–3 minutes · Per-arm sample and runtime · No signup
A/B Test Sample-Size & Duration Planner
Estimate the traffic needed to detect a chosen conversion-rate difference.
How the result works
Two-sided normal approximation for two independent proportions, 95% confidence, 80% power and equal allocation. Sample size is rounded upward; duration uses eligible traffic split equally between arms.
Method reviewed 2026-10-19 by Artyom Sklyarov, SUUR.
A worked example
A 10% baseline and 12% target need roughly 3,841 visitors per arm. At 1,000 eligible visitors daily the nominal runtime is eight days; account for full business cycles too.
What to check next
Fixed-horizon planning only: not a significance calculator or a sequential test. Rare conversions, dependence and multiple comparisons need specialist treatment.