Start with representative data
Use traffic and conversion rates from the exact audience, device mix, and entry point included in the experiment.
Free CRO tool
A/B test duration calculator
Estimate the sample size and planning window your experiment needs before you start sending traffic.
How to read the estimate
A planning tool, not a stopping rule.
The calculator uses a fixed-horizon, two-sided test for conversion rates. Treat the result as the minimum plan, then check data quality before declaring a winner.
Run through full business cycles
Keep tests live for at least two complete weeks and avoid launching across unusual promotions or outages.
Protect the decision
Review guardrail metrics, sample ratio mismatch, implementation quality, and practical impact alongside significance.
Common questions
Before you launch
Is uplift relative or absolute?
It is relative. A 10% uplift on a 5% baseline means the calculator looks for a 5.5% conversion rate.
Why is the recommendation at least 14 days?
Two full weekly cycles reduce the risk that weekday and weekend behavior distort the result, even when traffic is high.
Can I stop as soon as significance appears?
Not with this fixed-horizon estimate. Repeatedly checking and stopping early increases false positives unless your platform uses a valid sequential method.
Need help designing a trustworthy experiment?
Turn the estimate into a focused hypothesis, measurement plan, and test design.
Discuss an experiment →