A/B Test Duration Calculator
How long should you run an A/B test? The answer depends on your daily traffic, baseline conversion rate, and the size of improvement you want to detect. This calculator gives you a specific end date — not a vague "run until significant." Commit to the calculated duration before you start. Stopping early is the single most common cause of false positives in A/B testing.
A/B Test Duration Calculator
Find out how long you need to run your test to get reliable results
How to Use This Calculator
- Baseline conversion rate — your current CVR on the page you're testing. Pull from GA4 or your A/B testing tool for the last 30 days.
- Minimum detectable effect — the relative improvement you want to detect. 15–20% is a practical default. Lower MDE = longer required test duration.
- Daily visitors to the page — not site-wide traffic. Use the specific page's daily average from your analytics. Divide monthly sessions by 30 for a quick estimate.
- Number of variants — 2 for a standard A/B test (control + one variant). More variants require proportionally more traffic and time.
The Two Rules for Test Duration
A valid A/B test must satisfy both of these conditions before you declare a result:
- Minimum sample size reached — each variant has received the number of visitors calculated by the Sample Size Calculator.
- Minimum 14 days elapsed — always, regardless of sample size. Weekend/weekday behavioral differences make sub-two-week tests unreliable.
If your calculator result shows fewer than 14 days, the tool automatically raises it to 14. If it shows more than 90 days, consider increasing your MDE or testing on a higher-traffic page.
Frequently Asked Questions
How long should an A/B test run?
Minimum 14 days AND until the pre-calculated sample size is reached — whichever is longer. Stopping early inflates false positive rates from 5% to 26%+. Calculate your end date before starting, set a calendar reminder, and do not look at results until then.
What is the peeking problem?
Checking results before the end date and stopping when they look significant. A 95% confidence test that is peeked at repeatedly can produce false positives more than 26% of the time — 5× the intended rate. Fix: commit to the end date calculated before the test starts.
Can a test run too long?
Yes — tests over 90 days risk seasonal contamination. If you need more than 90 days, either increase your MDE or run the test on a higher-traffic page. Better to detect a 20% improvement confidently than to chase a 5% improvement across 4 months.
Why must tests run for at least 2 weeks?
Most sites have meaningfully different behavior on weekdays vs. weekends. A test that runs only 5–7 days captures an unrepresentative slice of your audience. 14 days ensures both weekday and weekend patterns are captured in both variants equally.
Not Sure What to Test First?
A free CRO audit identifies your highest-leverage test candidates — pages with the most traffic, worst CVR, and clearest hypotheses — so you spend your testing budget where it counts.
Get a Free CRO Audit