A/B Test Sample Size Calculator (100% FREE)
This a/b test sample size calculator shows roughly how much traffic your experiment needs. Enter your current conversion rate and the smallest improvement worth detecting to get an estimated sample size for each variation, with the assumptions shown.
Know roughly how much traffic your experiment needs.
Free, no signup. Runs in your browser, nothing is sent anywhere.
31,234
| Control | 31,234 |
|---|---|
| Variant | 31,234 |
| Total | 62,468 |
| Baseline → expected variant rate | 5.00% → 5.50% |
| Power, significance | 80%, 5% two sided |
A planning estimate, not a guarantee of a significant result. Decide the sample size up front and don't stop early when results look good: repeated peeking inflates false positives.
How it works
How to use the A/B Test Sample Size Calculator in three steps.
- Step 1
Enter your conversion rate
Enter the conversion rate you see today.
- Step 2
Set the smallest lift
Set the smallest relative uplift worth finding.
- Step 3
Review the estimate
Check the visitor estimate before you launch the test.
Examples
Sample sizes, set up properly.
Illustrative inputs. The calculator works out the actual numbers when you enter them.
A typical signup page
Input: Baseline 5%, relative MDE 10%, power 80%, two sided alpha 0.05. A 10% relative lift on 5% is half a percentage point.
ResultTarget variant rate 5.5%, with the visitors needed per variant calculated for you.
Bigger effect, smaller test
Input: Baseline 5%, relative MDE 30%. Bigger effects are easier to detect.
ResultFar fewer visitors per variant than at 10%.
How long it will run
Input: The same test with 2,000 eligible visitors a day. Helps decide if the test is realistic.
ResultAn estimate of days needed, based on the total sample.
Good to know
Planning a test.
What is minimum detectable effect?
MDE is the smallest difference a test is planned to detect with the chosen confidence and power. At a 5% baseline, a 10% relative lift means 5.5% conversion, which is 0.5 percentage points.
Sample size is a planning estimate
The calculation assumes independent visitors and a simple two arm test. Visitor quality, repeated tests, traffic changes, multiple comparisons and poor tracking can make real tests harder than the estimate suggests.
Method
Uses the standard two sided z test approximation for two independent proportions, solved for the sample per variant and rounded up.
FAQ
Questions, answered.
What is an A/B test sample size?
The approximate number of visitors needed in each variant to detect a planned difference in conversion under your chosen assumptions.
What is statistical power?
The chance the test detects the effect if it really exists. 80% is a common planning setting.
What does 95% confidence mean here?
A two sided significance level of 0.05. It isn't a 95% chance that your idea is right.
What is relative versus absolute MDE?
At a 5% baseline, a 10% relative lift is 5.5%. A 10 percentage point absolute lift would be 15%.
Can I stop as soon as the result looks significant?
Checking repeatedly and stopping early can invalidate standard significance results unless you use a method built for it.
Can this guarantee I'll see an improvement?
No. It estimates how big a test needs to be, not whether the effect exists.
Want tests that move numbers?
I help SaaS teams run experiments that lead to real growth.
See growth consulting