A/B Testing Memory Requirement Calculator Examples
Review worked A/B testing sample-size examples for different conversion rates, uplifts, traffic volumes, and test designs.
These examples show how baseline conversion rate, minimum detectable uplift, confidence, power, number of versions, and daily traffic affect the visitor requirement for an A/B test. Values are planning estimates based on equal traffic allocation.
Low-volume test with a 5% conversion baseline
A product team has 1,000 eligible visitors per day and wants to detect whether a new signup page raises conversion from 5.0% to 5.5%.
Input Summary
Baseline conversion rate
5%
Minimum detectable uplift
10%
Confidence and power
95% confidence, 80% power
Number of versions
2
Daily eligible visitors
1,000
Calculation Breakdown
- 1Absolute effect5% × 10%0.5 percentage points
- 2Visitors per versionTwo-proportion sample-size estimate31,243 visitors
- 3Total traffic31,243 × 262,486 visitors
- 4Estimated durationceil(62,486 / 1,000)63 days
Result Summary
Estimated duration
63 days
A/B Testing Memory Requirement Calculator
The test needs about 31,243 visitors in each version and approximately 63 days at the stated traffic level.
Higher baseline with a larger target change
An ecommerce team wants to measure whether a streamlined checkout can increase conversion from 20% to 24% using 10,000 eligible visitors per day.
Input Summary
Baseline conversion rate
20%
Minimum detectable uplift
20%
Confidence and power
95% confidence, 80% power
Number of versions
2
Daily eligible visitors
10,000
Calculation Breakdown
- 1Absolute effect20% × 20%4 percentage points
- 2Visitors per versionTwo-proportion sample-size estimate1,682 visitors
- 3Total traffic1,682 × 23,364 visitors
- 4Estimated durationceil(3,364 / 10,000)1 day
Result Summary
Estimated duration
1 day
A/B Testing Memory Requirement Calculator
The statistical traffic estimate is about 1,682 visitors per version, or 3,364 in total.
Three-version test with a low conversion rate
A B2B site converts 2% of eligible visitors and wants to detect a 25% relative uplift, from 2.0% to 2.5%, at 95% confidence and 90% power.
Input Summary
Baseline conversion rate
2%
Minimum detectable uplift
25%
Confidence and power
95% confidence, 90% power
Number of versions
3
Daily eligible visitors
3,000
Calculation Breakdown
- 1Absolute effect2% × 25%0.5 percentage points
- 2Visitors per versionTwo-proportion sample-size estimate16,873 visitors
- 3Total traffic16,873 × 350,619 visitors
- 4Estimated durationceil(50,619 / 3,000)17 days
Result Summary
Estimated duration
17 days
A/B Testing Memory Requirement Calculator
This three-version experiment needs approximately 50,619 eligible visitors overall and about 17 days of traffic.
How to Read Your Results
Visitors needed per version is the planned traffic target for every version, including the control.
Total visitors needed combines the planned samples for all equally allocated versions.
Estimated duration divides total eligible traffic needed by average daily eligible visitors and rounds up to whole days.
Expected conversions are an approximate count based on the average expected conversion rate, not a guaranteed outcome.
A short traffic estimate does not necessarily capture a representative mix of weekdays, campaigns, or customer behavior.
Assumptions & Important Notes
- All examples use equal traffic allocation across versions.
- The two-proportion approximation is used for the visitor estimates.
- Rates and traffic are assumed to remain broadly stable during each test.
- The calculation does not apply adjustments for multiple testing or repeated interim checking.
Related Examples
Frequently Asked Questions
Can an A/B test finish in one day if the calculator says one day?
It can reach the calculated traffic target in one day, but teams may also consider whether that day represents normal traffic behavior and whether the test plan requires a longer observation window.
Why does the three-version example need more total visitors?
Every version requires its own per-version sample under equal allocation. Adding a second variation changes the total from two planned samples to three.
What happens if daily eligible traffic falls below the estimate?
The estimated duration increases because it takes longer to collect the same total sample.
Are the examples suitable for revenue or engagement metrics?
They are most directly suited to binary conversion outcomes. Other metrics may need a different statistical approach.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.