A/B test significance calculator — know when a result is real, and how much traffic you need.
100% client-side
—
—
Confidence
⏳
Waiting for data
Enter visitors and conversions for both variants.
p-value
—
Rate A
—
—
Rate B
—
—
Absolute diff
—
percentage points
Relative lift
—
vs control
z-score
—
pooled z-test
Conf. interval
—
for the diff
A
—
B
—
95% confidence interval for the difference — does it cross zero?
—
current rate, %
significance level
chance of detecting a real effect
to estimate test duration
Sample needed per variant
—
visitors each
Total sample
—
both variants combined
Variant B target rate
—
—
Est. duration
—
at your weekly traffic
Tip: smaller effects need dramatically more traffic — a 1pp improvement on a 10% baseline takes ~4× the sample of a 2pp lift. If the required sample is far beyond your traffic, test a bigger change instead.