Calculate statistical significance for your conversion optimization tests.
Enter your test data above to see if your results are statistically significant.
This tool is free to use
If it saved you some time, consider buying me a coffee to support future updates.
Buy me a coffeeA/B Winner runs a two-proportion z-test on your test data — the standard statistical method for comparing conversion rates between two variants — and tells you whether Variant B's result is a genuine difference or could plausibly be random noise. Enter visitors and conversions for both variants, and it calculates confidence level, observed uplift, and p-value in real time as you type.
The verdict banner gives you a plain-English read: a clear winner or loser at 95%+ confidence, a "keep testing" signal between 80–95%, or "no significant difference" below that. The colour-coded confidence meter gives a quick visual sense of how close you are to a reliable result.
This is built for anyone running CRO or landing page tests — the goal is to stop teams calling a test "won" off a handful of extra conversions that don't hold up statistically.
Yes, completely free with no account or usage limit.
The p-value is the probability of seeing a difference this large (or larger) if there were genuinely no difference between the variants — a smaller p-value means the result is less likely to be random chance. Confidence level is simply 1 minus the p-value, expressed as a percentage; 95% confidence is the conventional threshold for calling a test statistically significant.
A two-proportion z-test, comparing the conversion rate of Variant B against Variant A using a pooled proportion and standard error calculation — the standard approach for comparing two conversion rates.
There's no fixed number — it depends on your baseline conversion rate and the size of the effect you're trying to detect. As a rule of thumb, be cautious of calling a winner on small sample sizes (a few hundred visitors per variant) even if confidence briefly crosses 95%, since results can swing significantly as more data comes in.
No. Every calculation runs locally in your browser using plain JavaScript — nothing is sent to a server.