Skip to content
Free tool

A/B test statistical significance calculator

Paste each variant's visitors and conversions to get the uplift, the p-value, and whether the result is real or noise.

Test results

Control (A)

Variant (B)

Chance variant B beats control (Spar)

99.5%

Spar's Bayesian read on a +24% relative uplift. The old frequentist verdict is significant at 95% (p = 0.0101).

Control rate

2.5%

Variant rate

3.1%

Relative uplift

+24%

Frequentist p-value

0.0101

How significance is calculated

This calculator runs a two-proportion z-test, the standard check for whether two conversion rates differ by more than chance. It pools the two variants to estimate the shared rate, measures how many standard errors apart the observed rates are, and converts that into a two-sided p-value.

z = (pB − pA) / √(p̂(1 − p̂)(1/nA + 1/nB))

The result is significant when the p-value falls below your threshold (0.05 for 95% confidence). A significant result means the difference is probably real; it does not, by itself, mean the lift is large enough to be worth shipping.

Read significance alongside effect size

Significance and size answer different questions. The p-value tells you whether a difference is real, and the uplift tells you whether it matters. Ship a change when both line up: a real difference that is also big enough to move revenue. Before you trust a flat result, confirm the test was powered to detect a meaningful lift, because an underpowered test that ends inconclusive is not the same as a tie.

FAQ

Frequently asked questions

Stop guessing what to test

A calculator tells you whether a test is worth running. Spar finds the tests worth running: it audits your storefront, ranks the gaps by revenue impact, and drafts the variations. Start a 14-day free trial.