Free tool
A/B test sample size.
Before launching a split test, work out how many visitors per variant you actually need to call a winner with confidence. Under-powered tests are worse than no tests.
Calculator
A/B test sample size
Required sample
- Per variant
- 53,212
- Target variant CVR
- 3.30%
How to use this
Baseline CVR is your current conversion rate on the page being tested. MDE is the smallest relative improvement you want to be able to detect. A 10 percent MDE on a 3 percent baseline means you want to catch a change from 3.0 to 3.3 percent.
Leave power at 80 and significance at 95 unless you have a strong reason to move them. Divide the required sample by your current weekly traffic to estimate how many weeks the test will need to run.
How we calculate
We use the standard two-proportion sample size formula: n per variant equals the quantity Z-alpha-over-two plus Z-beta, squared, multiplied by two p-bar times one minus p-bar, divided by delta squared. Here p-bar is the averaged conversion rate across control and variant, and delta is the absolute difference in conversion rate.
The Z values come from the normal distribution: roughly 1.96 for a two-tailed 95 percent significance level and 0.84 for 80 percent power.
Limitations
This is a frequentist sample size estimate. It assumes you run the test to the planned sample size and only peek once. Stopping early when the test looks significant inflates false positives.
For metrics that are not simple yes-no conversions (revenue per visitor, AOV), a proportion-based calculator under-sizes the test. Use a continuous-metric calculator or a sequential method for those cases.
Keep going
- CRO service for a full testing program.
- PDP grader to find the highest-value test ideas.
- Checkout friction audit for low-risk wins worth testing.