A/B Test Sample-Size Calculator · Minkyu Sung
Tools I built · Experimentation

A/B Test Sample-Size Calculator.

Answers the question every experiment starts with: how many people do we need? Enter your baseline rate, the effect you want to detect, power, and significance. It returns the required sample per variant and how long the test will run at your traffic.

Test parameters

Test type
Required sample
{{ nPer }}
per variant
{{ nTotal }}
Total across both
{{ duration }}
Weeks to run

Reading the result

To detect a lift from {{ crLabel }} to {{ variantRate }} with {{ powerLabel }} power at a {{ alphaLabel }} significance level, run {{ nPer }} users through each arm.

At {{ visitorsLabel }} visitors per week, that's about {{ duration }} weeks of runtime. Plan for full business cycles so day-of-week effects don't bias the read.

Two-proportion z-test Originally built in R Shiny Now native to this site