Baseline conversion
Use a recent conversion rate for the exact audience, page, and success event you intend to test. A borrowed industry benchmark can produce a sample estimate that looks precise but does not describe your experiment.
Free calculator
A/B Test Sample Size Calculator
Estimate the visitors required per variation using baseline conversion rate, minimum detectable lift, 95% confidence and 80% power.
Planning approximation for a two-sided two-proportion test using 95% confidence and 80% power. Validate important experiments with your analytics/statistics workflow.
Estimated requirement
17,943
visitors per variation
Traffic requirement too high?
Narrow pre-launch choices with a preference test, then validate behavioral lift once traffic supports it.
Create a preference testMethod and interpretation
How to plan an A/B test sample
Planning formula
n ≈ [(zα/2 × √(2p̄(1−p̄)) + zβ × √(p₁(1−p₁)+p₂(1−p₂))) ÷ (p₂−p₁)]²Minimum detectable lift
Enter the smallest relative improvement that would change the business decision. Smaller detectable effects require substantially more traffic; choosing an unrealistically large lift only makes the required sample look easier to reach.
Confidence and power
This planning estimate uses a two-sided 5% significance level and 80% power. That is a common starting point, not a universal rule. Higher power or a stricter false-positive threshold increases the required sample.
Worked example
If the baseline conversion rate is 4% and the smallest useful relative lift is 15%, the challenger rate is 4.6%. The calculator estimates the observations needed in each variation before the test starts; it does not declare a winner from observed results.