Experiment planning

A/B Test Sample Size Calculator

Estimate the traffic required for a fixed-horizon, two-sided test of two independent conversion rates with equal allocation.

Experiment assumptions

%
Current control-group conversion rate.
% relative
A 10% relative lift on a 10% baseline means an 11% target.
Total eligible visitors per day across both groups.
Estimated requirement

— users per group

Total sample
—
Control → target
—
Absolute effect
—
Estimated duration
Add daily traffic

This is a planning estimate, rounded up to whole users. It is not a guarantee of significance.

Method and limitations

The calculator uses the normal approximation for two independent proportions, a two-sided alpha, equal group sizes, and the requested power. It does not adjust for repeated peeking, multiple comparisons, clustering, sample-ratio mismatch, attrition, or variance reduction. Very low event rates and small samples may require simulation or an exact method.

Minimum detectable effect is entered as a relative lift. Always define the metric, analysis window, exposure unit, guardrails, and stopping rule before launch.

For the derivation, code examples, and experiment-planning workflow, read A/B Test Sample Size in Python and the broader A/B Test Engineering Guide.