A/B Test Sample Size Calculator
Work out how many visitors each variant needs before you launch, so you never ship a false winner.
Assumes 80% statistical power.
Stop calculating. Start testing.
Drop in your site and Dalton finds the experiment, sizes it, builds it, and ships the winner for you.
How to calculate sample size
Sample size depends on three things: your baseline conversion rate, the minimum effect you want to detect, and how confident you want to be. The calculator combines them with the standard two-proportion formula and returns the visitors needed per variant. As a rule of thumb, halving the effect you want to catch roughly quadruples the traffic you need, which is why low-traffic pages struggle to test small changes.
Why sample size matters
Calling a winner before you have enough data is the most common A/B testing mistake. Too small a sample and your result is noise dressed up as a win. Sizing the test up front tells you whether it is even feasible with your traffic, and how long it will take. Once results are in, use the statistical significance calculator to confirm the result is real.
Frequently asked questions
How do I calculate sample size for an A/B test?
Enter your baseline conversion rate and the minimum detectable effect you want to catch, then choose your significance and power. The calculator returns the visitors needed per variant using the standard two-proportion formula.
What is a good sample size for an A/B test?
There is no single number. It depends on your baseline conversion rate and the effect size you want to detect. Lower conversion rates and smaller effects need more visitors. The calculator gives you the exact figure for your case.
What is the minimum detectable effect (MDE)?
The smallest improvement you want the test to reliably catch. A relative MDE of 10 percent on a 3 percent baseline means detecting a lift to 3.3 percent. Smaller MDEs need much larger samples.
How does significance and power affect sample size?
Higher significance (fewer false positives) and higher power (fewer missed winners) both increase the sample you need. 95 percent significance and 80 percent power are the common defaults.
How long should I run the test?
Divide the total sample by your weekly traffic, and run for whole weeks to cover weekday and weekend behavior. The calculator estimates this if you enter your weekly visitors.