A/B Test Sample Size Calculator

A/B Test Sample Size Calculator: Two-proportion power analysis for experiments with a target lift, confidence, and power. It's built around ab testing, sample size, power analysis, minimum detectable effect, so you can go from a raw question to a checked answer without switching tools.

The math behind it: p₁ is the baseline conversion rate, p₂ = p₁ + minimum detectable effect, p̄ = (p₁ + p₂)/2, Z_{α/2} is tied to confidence, and Z_{β} represents statistical power. The core relationship is n = [Z_{α/2}√(2p̄(1 − p̄)) + Z_{β}√(p₁(1 − p₁) + p₂(1 − p₂))]² / (p₂ − p₁)², shown above the calculator so you can see exactly how your inputs turn into the result.

To use it well: (1) Enter your current conversion rate (control) and the absolute lift you want to be able to detect. (2) Choose a confidence level (alpha) and statistical power (1 − beta). (3) Optional: provide daily visitors to estimate how long the test will need to gather that many samples. Keep your units consistent as you go, and re-run a case you already know the answer to — it's the fastest way to catch a typo before it throws off a result you're relying on.

Worked example: entering Baseline = 3.5%, Lift = 0.5 pts, Confidence = 95%, Power = 80% returns ≈22,663 visitors per variant (≈45,326 total) to detect the lift reliably.. Try swapping in your own numbers next, especially a case you're unsure about, before you use this for something that matters.