A/B Test Sample Size Calculator
Find out how many visitors each variation needs and how long your test will run. Enter your baseline conversion rate and the smallest improvement worth detecting — the calculator does the rest.
Two-sided test, comparing one control against one variant. Run for at least 1–2 full business cycles (7–14 days) even if you hit the visitor count sooner.
How A/B test sample size is calculated
Sample size depends on four inputs: your baseline conversion rate, the minimum detectable effect (MDE)— the smallest lift you care about — the significance level (usually 95%, the tolerance for false positives), and statistical power (usually 80%, the chance of detecting a real effect). Smaller effects and lower baseline rates require dramatically more traffic.
The calculator uses the standard two-proportion formula for a two-sided test. It returns the visitors needed per variation; total traffic is roughly double that for a basic A/B test, and duration is the total divided by your daily traffic.
How to use the result
- If the duration is more than ~4 weeks, raise your MDE or accept you can’t reliably detect small wins — see A/B testing on low-traffic sites.
- Always run at least 1–2 full business cycles, even if you hit the count early — how long to run an A/B test.
- Don’t stop the moment you hit 95% — that’s peeking, and it inflates false positives. See the statistics mistakes guide.
Related reading
Not sure the test is worth running?
Predict the winner first. Get a Run/Iterate/Kill verdict before you spend the traffic.
Try AI prediction