How many visitors before a result means anything
Most tests on a small site stop long before they could have said anything. Enter your conversion rate, the smallest improvement you would act on and your daily traffic, and see how long an honest answer takes.
Visitors to measure the rate itself
750
Interval narrower than half the rate
Visitors per version, for an A/B test
13,807
Both versions together
27,614
Days at your traffic
139
What this means
At 200 visitors a day this test takes 139 days. Test a bigger change — a larger lift needs far fewer visitors.
How it works
Measuring a rate on its own: roughly 15 ÷ rate visitors puts the 95% interval within half the rate either side.
Comparing two versions: the standard two-proportion sample size at 95% confidence and 80% power. Halving the lift you want to detect roughly quadruples the visitors you need, which is why small sites should test big changes.