The finale
A/B test sample size calculator
How many users do you need for your experiment?
A quick recap of what shapes the sample size:
This is a pre-experiment calculator. It helps you plan how many visitors you need before you launch, so your results have a fair shot at clearing your chosen threshold.
Output
Minimum visitors in A
11,604
Minimum visitors in B
11,604
Each side needs ~11,604 visitors to detect a 10% relative lift on a 10% baseline at 95% confidence.
- False positive — you declare B a winner even if no real difference exists
- False negative — you declare A your winner, even if B is better
- Curve widths are kept constant to visualize the bells drifting apart, not scaled to calculated sample size.
How long will it take?
~24 days
At 1,000 visitors per day, you need ~24 days to collect the 23,208 total visitors required.
Assumptions
- Conversion rate metric. The numbers here assume you're measuring a conversion rate (e.g. signup, purchase) — not revenue, time-on-page, or other continuous metrics.
- One-tailed test. This calculator only checks whether B beats A — not whether A beats B. One tail, one direction.
- Relative effect. The minimum detectable effect (lift) input is a relative change (e.g. 10% lift means B is 10% better than A) rather than an absolute change (which would mean B is 10 percentage points better than A).
- Power fixed at 80%. Power is the chance of catching a real win if one exists. We lock it at 80% here and leave that lever for a later version of the calculator.
Set your baseline conversion, the smallest effect worth detecting, and your confidence level. The visitors-per-variant number tells you how many people each side of your test needs before you have a good chance to tell A from B.
Halve the lift and the required visitors roughly quadruple. Push confidence from 95% to 99% and the threshold slides further out, so you need more visitors to clear it. The relationships you've been reading off the chart now have numbers attached.
About the author
As a Product Manager and builder, I found that most A/B testing resources are too complex for beginners. I created this interactive guide and calculator to provide a simple, fun alternative. I hope it helps you get started!