Marketinga/b test durationtest runtimeexperiment duration

A/B Test Duration Calculator

Enter your daily traffic, baseline conversion rate, and minimum detectable effect. The calculator tells you how long to run your test before you can trust the results — preventing both early stopping and wasted runtime.

Advertisement

Calculator

%
%

See your A/B Test Duration Calculator results

Enter your email to unlock results — free forever.

or

No spam, ever. Unsubscribe at any time.

Advertisement

Formula

Duration = Required Sample Size × 2 ÷ Daily Visitors

The required sample size per variation is calculated using the two-proportion z-test formula. Multiplied by 2 for both variations, then divided by daily visitors (assuming a 50/50 traffic split) to get days.

How to use the A/B Test Duration Calculator

  1. 1

    Enter your average daily visitors

    Total daily visitors to the page being tested.

  2. 2

    Enter your baseline conversion rate

    Value should be in %.

  3. 3

    Enter your minimum detectable effect

    Smallest relative improvement worth detecting.

  4. 4

    Enter your statistical significance

  5. 5

    Read your results instantly

    Results update in real time as you type.

Advertisement

The cost of stopping too early

Peeking at test results and stopping early when you see significance is called 'optional stopping' — and it's statistically invalid. If you check results daily and stop at the first significant result, your actual false positive rate balloons from 5% to over 30%. You'll declare winners that are pure noise.

The fix is simple: calculate duration before you start, then don't look at results until you hit it. Many A/B testing platforms (Optimizely, VWO, Google Optimize) display results continuously, which makes this psychologically hard but no less important.

Why run for at least one full week

Even if your sample size is reached in 3 days, you should run for at least 7 days. Conversion rates vary significantly by day of week — weekday vs. weekend behavior, Monday intent vs. Friday browsing. A test that runs only Wednesday to Friday will skew toward that audience's behavior.

For most businesses, running for 2 full weeks gives you one cycle of every day type, plus enough data to average out holidays, promotions, and other one-time events.

Tips & Insights

Don't change the test mid-run

Any site change during the test period can contaminate results. Freeze all other changes to the tested page for the duration.

Include weekends

If your business has different weekend traffic patterns, always include at least two weekends in your test window.

Worked Examples

Mid-traffic SaaS landing page

Daily visitors: 800Baseline rate: 4%MDE: 20%Significance: 95%

About 19 days. At 800 visitors/day split 50/50, you're collecting 400 per variation per day and need ~7,600 total per variation.

Advertisement

Frequently Asked Questions

What if my traffic is highly seasonal?

Avoid running tests during atypical traffic periods (Black Friday, major sales events) unless the test is specifically for those events. Results from anomalous periods don't generalize to normal traffic behavior.

Can I extend a test that hasn't reached significance?

Yes, but you should set a maximum runtime in advance. A test that never reaches significance after 3× the expected duration likely has a true effect smaller than your MDE — meaning the variant probably isn't worth shipping.

Advertisement

Related Calculators