A/B Testing Screen Resolution Annual Formula
Learn how annual screen-resolution traffic, sample size, and test capacity are estimated for an A/B conversion test.
This calculation estimates whether visitors using a selected screen resolution provide enough annual traffic for a conversion experiment. It combines segment traffic, test allocation, baseline conversion rate, desired relative effect, and statistical settings to estimate the visitors required for one test.
- 100% Free
- No Sign-Up Required
- Private & Secure
- Mobile Friendly
Estimated Annual Tests Supported
Where:
First estimate the visitors in the chosen resolution segment who can enter tests. Then divide that traffic by the total sample required across every variant, rounding down to whole non-overlapping tests.
Variables Explained
| Variable | What It Means | Unit |
|---|---|---|
| annualVisitors - Annual visitors | Eligible visitors or sessions received over one year. | visits |
| resolutionShare - Screen resolution traffic share | Percentage of annual traffic using the selected resolution or resolution group. | percent |
| testTrafficAllocation - Traffic allocated to testing | Percentage of the selected resolution segment that can enter the experiment. | percent |
| baselineRate - Baseline conversion rate | Current segment conversion rate expressed as a decimal. | N/A |
| absoluteEffect - Absolute detectable effect | The desired relative effect converted into a decimal conversion-rate change. | N/A |
| variants - Number of variants | Total experiment variants, including the control. | number |
| confidenceZScore - Confidence z-score | Z-score representing the selected two-sided confidence level. | N/A |
| powerZScore - Power z-score | Z-score representing the selected statistical power. | N/A |
Step-by-Step Calculation
Estimate resolution-specific traffic
This isolates the annual visitors using the screen resolution or resolution group being assessed.
annualResolutionVisitors = annualVisitors * (resolutionShare / 100)
Find annual testable traffic
Only the portion of segment traffic available for experimentation is counted.
annualTestableVisitors = annualResolutionVisitors * (testTrafficAllocation / 100)
Convert the baseline rate
The percentage conversion rate is converted to a decimal for the sample-size calculation.
baselineRate = baselineConversionRate / 100
Calculate the absolute effect
A relative minimum detectable effect is translated into an absolute change in conversion rate.
absoluteEffect = baselineRate * (minimumDetectableEffect / 100)
Estimate sample per variant
This approximation estimates the visitors needed in each variant for a two-sided conversion-rate comparison.
sampleSizePerVariant = round((2 * baselineRate * (1 - baselineRate) * pow(confidenceZScore + powerZScore, 2)) / pow(absoluteEffect, 2))
Calculate the total test requirement
The per-variant requirement is multiplied by all variants, including the control.
totalRequiredVisitors = sampleSizePerVariant * variants
Estimate annual capacity
Available annual test traffic is divided by the traffic needed for one similarly sized test.
annualTestsSupported = floor(annualTestableVisitors / totalRequiredVisitors)
Example: a common desktop resolution segment
Resolution visitors
1,000,000 * (25 / 100)
250,000 visitors
Testable visitors
250,000 * (80 / 100)
200,000 visitors
Absolute effect
(4 / 100) * (20 / 100)
0.008 or 0.8 percentage points
Sample per variant
round((2 * 0.04 * 0.96 * pow(1.96 + 0.84, 2)) / pow(0.008, 2))
9,408 visitors
Total required visitors
9,408 * 2
18,816 visitors
Annual tests supported
floor(200,000 / 18,816)
10 tests
Final Result
About 200,000 testable visitors are available per year. One test requires about 18,816 visitors in total, using about 9.4% of annual testable traffic.
Assumptions
- ✓The selected resolution segment's traffic share remains broadly stable during the year.
- ✓Visitors are split evenly among variants.
- ✓The conversion metric is binary and the approximation is used for a two-sided comparison.
- ✓Tests use independent visitors and are treated as non-overlapping.
- ✓The chosen baseline rate and minimum detectable effect reasonably represent the planned experiment.
Limitations
- !Actual requirements can be higher when traffic allocation is uneven or variant allocation is not equal.
- !Repeat visitors, data loss, bot filtering, and cross-device behavior can reduce usable sample size.
- !Multiple metrics, repeated looks at results, or multiple comparisons are not included in this simplified estimate.
- !A resolution group may contain materially different devices or layouts, making a pooled result less representative.
- !Annual capacity does not guarantee that traffic arrives evenly enough to meet a desired test duration.
Common Mistakes to Avoid
Entering a percentage as a decimal, such as entering 0.25 instead of 25 for a 25% traffic share.
Treating a 20% relative effect from a 4% baseline as a 20-percentage-point change instead of a 0.8-percentage-point change.
Excluding the control when entering the total number of variants.
Using total site conversion rate rather than the baseline conversion rate for the selected resolution segment.
Assuming that all segment visitors are eligible when only part of the traffic can be exposed to the test.
Related Formulas
Frequently Asked Questions
How is A/B test sample size calculated for a screen resolution?
The estimate uses the segment baseline conversion rate, absolute detectable effect, confidence z-score, power z-score, and the number of variants. It then compares the total required sample with testable annual segment traffic.
Why does a smaller minimum detectable effect need more traffic?
A smaller effect is harder to distinguish from normal variation. Because the effect appears in the denominator squared, reducing it can increase the required sample substantially.
Does the required sample include the control group?
Yes. The sample per variant is multiplied by the total number of variants, including the control.
What does annual traffic coverage mean?
It is the percentage of annual testable segment traffic needed for one test. A result below 100% means one test fits within the annual traffic estimate.
Ready to calculate your result?
Use the calculator to get instant results with your own inputs.