Free tool
Conversion, A/B test and sample size calculator
Three calculators on one page: your conversion rate, whether variant B really beat A, and how many visitors you need before you start.
- Two-sided z-test
- Instant results
- Free, no sign-up
Conversion rate
Conversions divided by visitors.
Conversion rate
—
A/B test significance
Is the difference between two variants real, or noise? A two-sided test of two proportions.
- Uses a pooled two-proportion z-test
- Pick the confidence level before the test starts
- Stop at the sample size you planned, not the first time it looks significant
Variant A (control)
Variant B
Rate A
—
Rate B
—
Relative uplift
—
p-value
—
Sample size
How many visitors each variant needs to detect the change you care about.
- Smaller effects need far more visitors: halving the effect roughly quadruples the sample
- Power is the chance of detecting a real change of that size
- Run whole weeks so weekdays and weekends are both covered
Your current rate, e.g. 5
Smallest uplift worth detecting, e.g. 10
Visitors per variant
—
How to use it
Plan, run, decide
- 1
Size the test
Enter your baseline and the smallest uplift worth detecting.
- 2
Run until you reach it
Don’t stop early because one day looks good.
- 3
Check significance
Enter the results and read the verdict.
Tips
Avoid the common traps
Run full weeks
Weekdays and weekends convert differently. Run tests in whole weeks.
Don’t peek and stop
Stopping the moment a result looks significant inflates false positives.
Count humans only
Bots dilute conversion rates. PathClue reports conversion on human traffic.
FAQ
Calculator questions
Anything else — open the chat in the corner and ask.
Which test does the significance calculator use?
A two-sided z-test for the difference between two proportions, using the pooled conversion rate.
What does the p-value mean?
The chance of seeing a difference at least this large if the two variants actually performed the same. Below 0.05 is significant at 95% confidence.
Why is the sample size so large?
Small uplifts on low conversion rates are hard to tell apart from noise. Testing bigger changes, or a step with a higher rate, needs fewer visitors.