A/B Testing Core Web Vital Per-User Formula
Learn how this calculator measures Core Web Vital differences per user and estimates time, conversion, and value impact from an A/B test.
This calculator compares the measured Core Web Vital for a test variant with the control experience. It turns the observed difference into a per-user metric change, aggregate metric reduction across variant users, and an estimated commercial impact based on the observed conversion-rate difference.
- 100% Free
- No Sign-Up Required
- Private & Secure
- Mobile Friendly
Estimated Incremental Value
Where:
Find the percentage-point conversion-rate difference, apply it to the number of users who saw the variant, then multiply the estimated additional conversions by the value of one conversion.
Variables Explained
| Variable | What It Means | Unit |
|---|---|---|
| controlMetric - Control metric value | The measured LCP, INP, or scaled CLS value for the control experience. | milliseconds or scaled CLS score |
| variantMetric - Variant metric value | The measured value for the same Core Web Vital in the variant experience. | milliseconds or scaled CLS score |
| variantUsers - Users exposed to variant | The number of users who saw the variant during the measurement period. | users |
| controlConversionRate - Control conversion rate | The observed control conversion rate expressed as a percentage. | percent |
| variantConversionRate - Variant conversion rate | The observed variant conversion rate expressed as a percentage. | percent |
| valuePerConversion - Value per conversion | The revenue or assigned business value of one completed conversion. | currency |
Step-by-Step Calculation
Calculate the per-user metric change
For LCP and INP, a positive result means the variant has a lower and therefore faster metric value. For scaled CLS, a positive result means the variant has a lower layout-shift score.
perUserMetricChange = controlMetric - variantMetric
Calculate the percentage metric improvement
This expresses the metric reduction relative to the control value.
perUserImprovementPercent = (perUserMetricChange / controlMetric) * 100
Estimate aggregate metric reduction
The per-user difference is multiplied by variant users and converted from milliseconds to seconds. For CLS, this output is a scaled aggregate metric difference rather than literal time saved.
totalTimeSavedSeconds = (perUserMetricChange * variantUsers) / 1000
Calculate relative conversion-rate lift
This compares the conversion-rate change with the control conversion rate.
conversionRateLift = ((variantConversionRate - controlConversionRate) / controlConversionRate) * 100
Estimate incremental conversions
The absolute conversion-rate difference in percentage points is applied to variant users.
incrementalConversions = variantUsers * ((variantConversionRate - controlConversionRate) / 100)
Estimate incremental value
Estimated additional conversions are multiplied by the value assigned to each conversion.
incrementalValue = incrementalConversions * valuePerConversion
Worked example: LCP improvement in a checkout experiment
Per-user LCP improvement
3,000 ms - 2,400 ms
600 ms
LCP improvement percentage
(600 / 3,000) * 100
20.0%
Aggregate time reduction
(600 * 10,000) / 1,000
6,000 seconds
Conversion-rate lift
((3.2 - 3.0) / 3.0) * 100
6.7%
Estimated incremental conversions
10,000 * ((3.2 - 3.0) / 100)
20 conversions
Estimated incremental value
20 * 50
$1,000
Final Result
The variant shows a 600 ms per-user LCP improvement, a 6.7% observed conversion-rate lift, and an estimated $1,000 incremental value for the 10,000 variant users measured.
Assumptions
- ✓Control and variant metrics are collected using the same Core Web Vital definition, measurement method, time period, and comparable traffic mix.
- ✓Lower values represent improvement for LCP, INP, and scaled CLS inputs.
- ✓The average metric difference is applied uniformly across all users exposed to the variant.
- ✓The observed conversion-rate difference is used as an estimate of incremental conversions.
- ✓Value per conversion is consistent across the conversions included in the calculation.
Limitations
- !The calculator does not test statistical significance, confidence intervals, sample-ratio mismatch, or experiment validity.
- !An observed conversion difference does not prove that the Core Web Vital change caused the difference.
- !Aggregate time saved is most directly meaningful for LCP and INP; it should not be interpreted as literal user time saved when the selected metric is CLS.
- !Results from a test period may not persist with a different audience, season, traffic source, or implementation.
- !Estimated value excludes costs, refunds, margins, repeat behavior, and other business effects unless included in the entered value per conversion.
Common Mistakes to Avoid
Entering a control and variant metric from different tools, reporting periods, or device mixes.
Treating a lower conversion rate as a positive result because the performance metric improved.
Using relative conversion lift instead of the percentage-point difference when calculating incremental conversions.
Entering a CLS score such as 0.120 instead of the calculator's scaled value of 120.
Reading aggregate metric reduction as the exact amount of time every visitor personally saves.
Treating an observed test difference as conclusive without assessing experiment reliability.
Related Formulas
Frequently Asked Questions
How is per-user Core Web Vital improvement calculated?
Subtract the variant metric value from the control metric value. A positive result means the variant has a lower measured value.
How do I calculate Core Web Vital improvement as a percentage?
Divide the per-user metric change by the control metric value, then multiply by 100.
How are incremental conversions estimated?
The calculator multiplies variant users by the absolute conversion-rate difference in percentage points.
Why does conversion lift differ from the conversion-rate difference?
Conversion lift is relative to the control rate, while incremental conversions use the absolute percentage-point difference.
How should CLS be entered in this formula?
Use a consistent scaled value for both groups. For example, enter 120 for a CLS score of 0.120.
Can the calculated value be negative?
Yes. A negative result means the observed variant conversion rate was lower than the control rate, or the variant metric was worse.
Ready to calculate your result?
Use the calculator to get instant results with your own inputs.