Average daily visitorsThe number of visitors your website receives per dayBaseline conversion rate (%)Your current conversion rate before running the testMinimum detectable effect (%)The minimum relative change you want to detectStatistical significance (%)Confidence level for the test results

Estimated test duration

Updates as you type

17 days (~2 weeks)

8,150

per variation

16,300

total visitors

Why the test takes this long

Every day adds visitors, and every visitor makes the result less noisy. Move across the figure to step through the days and watch the two possible results pull apart. It uses the numbers in the calculator above.

A: controlB: variantEach ball is one possible result of the same test. Hopper A drops at your baseline (5.00%), hopper B at the lift you want to detect (6.00%); the pegs add the noise a finite sample really has. With few visitors the piles overlap and a real lift is easy to miss. At 8,150 visitors per version they sit far enough apart that 80% of tests would find it at 95% significance.

Understanding Test Duration

Sample Size Calculation: The calculator uses an industry-standard formula that accounts for both your baseline conversion rate and the minimum effect you want to detect.

Daily Traffic Split: The calculator assumes an equal 50/50 split of your daily traffic between the control and test variations.

Confidence Level: A higher confidence level (e.g., 99% vs. 95%) increases the required sample size. Industry standard is 95%.

Statistical Power: This calculator uses a fixed 80% power level, which is the standard for most A/B tests and balances the risk of false negatives.

Best Practices for A/B Testing

•Run the full duration: Even if you see promising results early, run the test for the full recommended duration to ensure statistical validity.
•Include full business cycles: Make sure your test duration includes full business cycles (e.g., weekdays and weekends) to account for traffic variations.
•Test one change at a time: To clearly understand what impacts your conversion rate, test one element at a time.
•Document everything: Keep detailed records of what you tested, why, and the results for future reference.

More for a/b testing

All 18 free tools

Sample sizeVisitors you need for a valid test A/B test significanceDid your variant really win Conversion rateYour rate against benchmarks

You know when to call it. Now launch it.

Humblytics handles traffic allocation, tracks statistical significance in real-time, and tells you exactly when your test reaches a valid result, no spreadsheets required.

Start free trial Book a demo

14-day trial · Card required to start · Cancel anytime