Marketing · Free Tool
A/B Test Significance & Sample Size Calculator
Check whether an A/B test result is statistically significant, or calculate the sample size you need before you launch one.
Control (A)
Variant (B)
Assumes 80% statistical power — the common default for A/B test sample-size planning.
About this tool
This calculator covers both ends of an A/B test. In "Check significance" mode, enter the visitors and conversions for your control and variant and it runs a two-proportion z-test, reporting each group's conversion rate, the relative uplift or decline, and a confidence level such as "97.3% confidence that the variant differs from control" — with a clear call on whether that clears the common 95% significance bar. In "Sample size" mode, enter your baseline conversion rate and the minimum relative lift you care about detecting, and it uses the standard two-proportion power formula (fixed at 80% power, the conventional default) to tell you how many visitors per variant — and in total — the test needs before you start.
Everything runs client-side and updates live as you type, so there is nothing to install and no data leaves your browser. The tool also protects against the two most common ways teams fool themselves with A/B tests: it flags results based on very small samples (under roughly 100 conversions or 1,000 visitors per group) as unreliable even if they look "significant," and in sample-size mode it will translate the required total into an estimated number of days if you provide your average daily traffic, so you know upfront whether a test is even feasible to run to completion.
Common uses
- Decide whether a completed A/B test result is real or just noise before rolling out a winning variant
- Calculate the minimum sample size and test duration needed before launching an experiment, so you don't stop it too early
- Sanity-check a "we saw a 20% lift!" claim from a small pilot test that may not hold up statistically
Frequently asked questions
What statistical test does this use?
A two-proportion z-test for the significance check, and the standard two-proportion sample-size formula (with 80% statistical power) for the sample-size calculator — the same math behind most dedicated A/B testing calculators.
Why does it warn me about small samples even when the result says "significant"?
A p-value can look significant on a tiny sample purely by chance. As a rule of thumb, results based on fewer than about 100 conversions or 1,000 visitors per group are noisy and prone to reversing as more data comes in, so treat them as directional, not final.
Why is "minimum detectable effect" relative rather than absolute?
A relative MDE (e.g. "detect at least a 10% lift") scales with your baseline rate, which is how most CRO teams plan tests — a 10% relative lift on a 5% baseline means detecting a move to 5.5%, not to 15%.
Is my data uploaded to a server?
No. This tool runs entirely in your browser using JavaScript — nothing you type is transmitted, logged, or stored anywhere. You can even use it offline once the page has loaded.