A/B Testing Sample Size (Monthly) Calculator Examples
Worked A/B testing sample size scenarios showing how conversion rate, uplift, traffic, power, and test eligibility affect test duration.
These examples illustrate how the same A/B testing method produces very different traffic requirements when the baseline conversion rate, minimum detectable uplift, confidence level, power, or eligible monthly traffic changes.
How to Read Your Results
Visitors per variant is the target eligible audience for the control and for the variant separately.
Total visitors needed is twice the per-variant figure because this calculator assumes an even two-version split.
Estimated months uses eligible test traffic, not all reported site visitors.
A small relative uplift can represent a very small absolute conversion-rate change, which typically needs more traffic.
Treat the duration as a planning estimate and allow time for setup, quality checks, and normal traffic variation.
Assumptions & Important Notes
- All examples assume one control and one variant with a 50/50 traffic allocation.
- Examples use a two-sided conversion-rate comparison.
- Traffic is assumed to arrive at a roughly consistent average monthly pace.
- The shown inputs are illustrative planning scenarios, not recommended settings for every experiment.
Related Examples
Frequently Asked Questions
What is an example of a low-traffic A/B test plan?
A low-traffic plan may require several months when the expected conversion difference is small. The relevant input is eligible monthly traffic, not all visits.
Why can a high-converting page still need a large sample?
If the test is designed to detect only a small relative uplift, the absolute difference may still be small and difficult to distinguish from random variation.
What happens if only 50% of visitors are eligible for a test?
Eligible monthly traffic is halved, so the estimated duration generally doubles if the required total sample stays the same.
How does 90% power change a worked example?
Using the 90% power z-score of 1.282 instead of 0.84 increases the calculated sample requirement, all else equal.
Can I use monthly averages for a seasonal website?
You can use them for a rough estimate, but a monthly average may be less representative when traffic or visitor behavior varies substantially by season.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.