Before the test: sample size per variant, duration in weeks and the smallest uplift your traffic can detect. After the test: significance, p-value, uplift and its confidence interval.
Before the test: enter your baseline conversion rate, the smallest relative uplift worth detecting, daily traffic entering the test, the number of variants, and your confidence and power levels. After the test: enter visitors and conversions for the control (A) and the variant (B).
"Significant" only says the uplift is probably not zero. A +17% result with an interval from +2% to +32% means you should plan on the low end. Most calculators give a yes or no; this one gives the range you can bet on.
If the test needs more than eight weeks, seasonality and cookie loss will weaken the result. Test bolder changes, test higher in the funnel where volumes are larger, or group similar pages into one test. The calculator shows the smallest uplift you can detect in four weeks with your current traffic.
Fix the sample size and the end date before launch and do not stop the test as soon as it turns green: peeking inflates false positives.
Get access to playbooks and feedback from those who are shaping the next wave of marketing and AI.
