Free tool · no signup

A/B test sample size.
How many users before you can call a winner.

Standard two-proportion z-test maths. Tell us your current conversion rate, the smallest improvement you care about (relative lift), and how strict you want to be. We'll tell you how many visitors per variant the test needs — and how long it'll run at your traffic level.

e.g. 5 = 5% of visitors currently convert.
Relative — 10 = you want to catch a 10% improvement (so 5.0% → 5.50%). Smaller lifts need exponentially more samples.
Half your total daily traffic (assuming 50/50 split). Lets us estimate how many days the test will run.
A note on assumptions: The formula assumes a 50/50 traffic split, independent observations, and a single primary metric. For multiple metrics or sequential testing, you need either a Bonferroni correction or a sequential-testing framework — talk to whoever runs analytics.
Per variant
31,240
visitors per variant · 62,480 total
Control (baseline)5.00%
Treatment (detectable)5.50%
Significance95% conf · 80% power

Vyrable's headlines A/B engine ships two variants of every LinkedIn post, picks a winner automatically, and applies the lesson to your next post. No manual stats math.