A/B Test Sample Size Calculator

How to use this tool

1. Enter the baseline conversion rate as a decimal (0.10 means 10%) and the smallest absolute lift you need to detect.
2. Pick confidence and power, then press Generate. The tool shows the visitors needed per variant.
3. Use Load sample for baseline 10%, effect 2 points, 95% confidence and 80% power, or Clear to start again.

Worked example

For baseline 0.10, minimum detectable effect 0.02, 95% confidence (z = 1.96) and 80% power (z = 0.84), the per-variant size is 2 x (1.96 + 0.84)^2 x 0.10 x 0.90 / 0.02^2, which is about 3,532 visitors per variant (rounded up).

What this tool does not do / Limits

This is a normal-approximation calculation for teaching and planning. It assumes a two-sided two-proportion test with equal split and independent visitors; it ignores peeking, segmentation, multiple variants and real-world variance, so treat the result as a planning figure, not a guarantee.

Frequently asked questions