Guides ยท Business

A/B Test Sample Size Basics

Plan tests with enough power

Planning A/B tests means estimating baseline metrics, choosing a meaningful detectable effect, and setting significance/power to compute required samples/duration, avoiding peeking and underpowered tests.

Define Inputs

Baseline, MDE, alpha, and power; choose primary metric.

Calculate

Use calculators or stats libs; adjust for variance and traffic.

Run Properly

Avoid peeking; ensure duration covers cycles/seasonality.

Related Terms