A/B Test Sample Size Calculator
Enter your current conversion rate and the smallest improvement you care about. You get the visitors needed for each version and the days it should take.
Worked examples
10% to 12%
Current rate 10% Improvement 2 percentage points 95% confidence, 80% power
3,841 visitors for each version, 7,682 in total
At 400 visitors a day, 7,682 visitors takes 20 days.
A low rate and a relative lift
Current rate 3% Improvement 20% of the current rate (3% to 3.6%)
13,914 visitors for each version
Low rates need much bigger samples to show the same relative change.
About A/B Test Sample Size Calculator
A test with too few visitors is mostly noise. You may see a big difference that disappears next week, or miss a real improvement because the sample was too small to show it. The sensible approach is to decide how many visitors you need before you start, and that number depends on how common conversions are and how small a change you want to be able to spot.
Enter your current conversion rate and the smallest improvement worth finding, either in percentage points or as a share of the current rate. The calculator uses the standard two-proportion sample size formula with your chosen confidence and power, and returns the visitors needed for each version and in total. Add how many visitors you get a day and it works out how many days the test needs. You will see quickly that small lifts on low rates need very large numbers, which is a good reason to test bold changes.
How to use A/B Test Sample Size Calculator
- Enter your current conversion rate as a percentage.
- Enter the smallest improvement you want to detect and say whether it is in percentage points or a share of the current rate.
- Pick your confidence and power. 95% and 80% are the usual choices.
- Add the daily visitors to the test to see how long it should run.
Common uses
- Planning a test before building it
- Deciding whether a page has enough traffic to test at all
- Setting expectations with a team about how long a test takes
- Choosing between a small tweak and a bold change
Frequently asked questions
What is power?
Power is the chance that the test will spot a real improvement of the size you chose. At 80%, a real lift of that size will be found four times in five. Higher power needs more visitors.
What is the smallest improvement I should pick?
The smallest one that would change a decision, for example one large enough to cover the cost of making the change. Asking to detect tiny lifts needs huge samples.
What is the difference between percentage points and relative?
Going from 10% to 12% is 2 percentage points, or a 20% relative lift on the current rate. The calculator accepts either, and shows the rate you are aiming to detect.
Why run for whole weeks?
Visitors behave differently on weekdays and weekends, so a test covering whole weeks gives a fairer picture than one that ends on a Tuesday.
What if I can not reach the number?
Test a bigger change, test on a page with more traffic, or accept a lower confidence or power knowing the result will be less certain.