Find out how many visitors each version of your A/B test needs before you can trust the result, and how long the test should run.
Loading the tool…
What you need to enter
- Current conversion rate: how your control performs today.
- Smallest lift to detect: the smallest relative improvement worth knowing about. Smaller lifts need far more visitors.
- Confidence: usually 95%, so there’s a 5% chance of a false winner.
- Power: usually 80%, the chance of spotting a real lift of that size.
How sample size changes
| Baseline rate | Lift to detect | Visitors per variant (95%, 80%) |
|---|---|---|
| 3% | 10% | about 53,000 |
| 3% | 20% | about 13,900 |
| 10% | 10% | about 14,800 |
| 10% | 20% | about 3,800 |
Rules for a fair test
- Decide the sample size before you start.
- Don’t stop the test early just because one version is ahead.
- Run for full weeks so every weekday is included.
- Change one thing at a time where you can.
Frequently asked questions
How many visitors do I need for an A/B test?
It depends on your conversion rate and the lift you want to detect. Small lifts on low conversion rates can need tens of thousands of visitors per variant.
How long should an A/B test run?
Until you reach the planned sample size, and at least one full week.
What is statistical power?
The chance your test detects a real difference of the size you chose. 80% is standard.
What if I don't have enough traffic?
Test bigger changes, use a higher-traffic page, or accept a lower confidence level.
Last updated: