A/B Testing Sample Size Calculator Examples
Worked A/B testing sample-size examples show how baseline conversion rate, target uplift, confidence, power, and traffic change a test plan.
These examples use an evenly split, two-variation conversion test with 95% confidence and 80% power unless stated otherwise. They illustrate why small effects and low conversion rates often need substantial traffic.
Landing page test with a 10% baseline conversion rate
Medium-traffic landing page conversion experiment
Input Summary
Baseline conversion rate
10%
Minimum detectable uplift
20%
Confidence and power
95% confidence, 80% power
Daily eligible visitors
1,000
Calculation Breakdown
- 1Target variant conversion rate10% * 1.2012%
- 2Visitors per variationTwo-proportion sample-size approximation3,837 visitors
- 3Total visitors3,837 * 27,674 visitors
- 4Estimated durationceil(7,674 / 1,000)8 days
Result Summary
Estimated duration
8 days
A/B Testing Sample Size Calculator
The test needs about 3,837 visitors per variation and 7,674 total visitors.
Low-conversion ecommerce purchase test
Low baseline purchase conversion with moderate traffic
Input Summary
Baseline conversion rate
2%
Minimum detectable uplift
20%
Confidence and power
95% confidence, 80% power
Daily eligible visitors
5,000
Calculation Breakdown
- 1Target variant conversion rate2% * 1.202.4%
- 2Visitors per variationTwo-proportion sample-size approximation21,481 visitors
- 3Total visitors21,481 * 242,962 visitors
- 4Estimated durationceil(42,962 / 5,000)9 days
Result Summary
Estimated duration
9 days
A/B Testing Sample Size Calculator
The test needs about 21,481 visitors per variation and about 9 days at 5,000 eligible visitors per day.
High-traffic signup test for a small uplift
Large audience, conservative minimum detectable effect
Input Summary
Baseline conversion rate
15%
Minimum detectable uplift
5%
Confidence and power
95% confidence, 80% power
Daily eligible visitors
20,000
Calculation Breakdown
- 1Target variant conversion rate15% * 1.0515.75%
- 2Visitors per variationTwo-proportion sample-size approximation43,989 visitors
- 3Total visitors43,989 * 287,978 visitors
- 4Estimated durationceil(87,978 / 20,000)5 days
Result Summary
Estimated duration
5 days
A/B Testing Sample Size Calculator
The test needs about 43,989 visitors per variation, or 87,978 visitors total.
Higher-certainty lead-generation test
Lead-generation test with stricter statistical settings
Input Summary
Baseline conversion rate
5%
Minimum detectable uplift
25%
Confidence and power
99% confidence, 90% power
Daily eligible visitors
800
Calculation Breakdown
- 1Target variant conversion rate5% * 1.256.25%
- 2Visitors per variationTwo-proportion sample-size approximation using Z-scores 2.576 and 1.2825,246 visitors
- 3Total visitors5,246 * 210,492 visitors
- 4Estimated durationceil(10,492 / 800)14 days
Result Summary
Estimated duration
14 days
A/B Testing Sample Size Calculator
The test needs about 5,246 visitors per variation and an estimated 14 days.
How to Read Your Results
Visitors per variation is the minimum planning estimate for both the control and variant groups.
Total visitors is the combined requirement across the two groups, assuming a 50/50 traffic allocation.
Estimated days divides total required visitors by daily eligible visitors and rounds up.
Target variant conversion rate shows the conversion rate implied by the relative uplift you selected.
Treat the output as a planning estimate rather than a promise that a test will produce a significant result.
Assumptions & Important Notes
- Every included visitor has a binary conversion opportunity and is independently assigned to a variation.
- Traffic is split evenly between two variations.
- The selected confidence and power use the calculator's displayed Z-score settings.
- Traffic quality and the baseline conversion rate are broadly stable during the test.
- Examples do not account for multiple variants, segmentation, or sequential testing rules.
Related Examples
Frequently Asked Questions
How many examples should I model before launching an A/B test?
Model at least a realistic target uplift and a conservative smaller uplift. Comparing both can show whether the available traffic supports the decision you need to make.
Why does the low-conversion example need more visitors?
Its 20% relative uplift produces only a 0.4-percentage-point absolute difference, which is harder to separate from normal variation.
Can I use these examples for email A/B tests?
They can illustrate binary outcomes such as opens or clicks, provided recipients are independently assigned and the metric definition is consistent.
Should I use calendar days or eligible visitors to plan length?
Use eligible visitors as the sample requirement. Calendar days are a traffic-based estimate and should also account for normal weekly or seasonal behavior.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.