A/B Testing Core Web Vital Per-User Calculator Examples
Worked examples show how Core Web Vital test results can be translated into per-user improvement, conversion differences, and estimated value.
These examples use the calculator's method to compare a control with a variant. They illustrate different metrics and outcomes, including a positive LCP change, an INP improvement with no conversion change, a scaled CLS comparison, and a case where a faster variant has lower observed conversions.
Ecommerce checkout LCP improvement
10,000 variant users see a checkout page with a faster LCP.
Input Summary
Core Web Vital
LCP
Control metric
3,000 ms
Variant metric
2,400 ms
Variant users
10,000
Control conversion rate
3.0%
Variant conversion rate
3.2%
Value per conversion
$50
Calculation Breakdown
- 1Metric change3,000 - 2,400600 ms improvement
- 2Metric improvement(600 / 3,000) * 10020.0%
- 3Incremental conversions10,000 * ((3.2 - 3.0) / 100)20
- 4Incremental value20 * 50$1,000
Result Summary
Incremental value
$1,000
A/B Testing Core Web Vital Per-User Calculator
The test shows a 600 ms LCP reduction and an estimated $1,000 incremental value in the measured variant audience.
SaaS sign-up INP improvement with flat conversion
25,000 users see a variant with lower interaction latency.
Input Summary
Core Web Vital
INP
Control metric
280 ms
Variant metric
210 ms
Variant users
25,000
Control conversion rate
5.0%
Variant conversion rate
5.0%
Value per conversion
$120
Calculation Breakdown
- 1INP change280 - 21070 ms improvement
- 2Aggregate time reduction(70 * 25,000) / 1,0001,750 seconds
- 3Conversion lift((5.0 - 5.0) / 5.0) * 1000.0%
- 4Incremental value25,000 * ((5.0 - 5.0) / 100) * 120$0
Result Summary
Incremental value
$0
A/B Testing Core Web Vital Per-User Calculator
The variant improves INP by 70 ms per user but has no observed conversion-rate change in this test period.
Content site CLS reduction
A publishing site tests reserved ad space to reduce layout shifts for 80,000 variant users.
Input Summary
Core Web Vital
CLS
Control metric
180 (CLS 0.180)
Variant metric
90 (CLS 0.090)
Variant users
80,000
Control conversion rate
1.5%
Variant conversion rate
1.6%
Value per conversion
$8
Calculation Breakdown
- 1Scaled CLS change180 - 9090 scaled points
- 2CLS improvement(90 / 180) * 10050.0%
- 3Incremental sign-ups80,000 * ((1.6 - 1.5) / 100)80
- 4Incremental value80 * 8$640
Result Summary
Incremental value
$640
A/B Testing Core Web Vital Per-User Calculator
The variant reduces the scaled CLS input by 90 points and corresponds to an estimated $640 value difference.
Faster landing page with lower observed conversions
A performance change is released alongside a visual redesign to 15,000 variant users.
Input Summary
Core Web Vital
LCP
Control metric
2,500 ms
Variant metric
2,100 ms
Variant users
15,000
Control conversion rate
4.0%
Variant conversion rate
3.8%
Value per conversion
$75
Calculation Breakdown
- 1LCP change2,500 - 2,100400 ms improvement
- 2LCP improvement percentage(400 / 2,500) * 10016.0%
- 3Estimated incremental conversions15,000 * ((3.8 - 4.0) / 100)-30
- 4Estimated incremental value-30 * 75-$2,250
Result Summary
Estimated incremental value
-$2,250
A/B Testing Core Web Vital Per-User Calculator
Despite a 400 ms LCP improvement, the variant has an estimated negative value difference of $2,250 in the measured group.
How to Read Your Results
A positive per-user metric change means the variant has a lower entered metric value than the control.
Metric improvement percentage puts the absolute change in context of the control baseline.
Conversion lift is relative; a move from 3.0% to 3.2% is a 0.2 percentage-point change and a 6.7% relative lift.
Incremental value is a test-period estimate based on the entered value per conversion, not a guaranteed future result.
For CLS, use the metric comparison outputs but do not treat aggregate scaled CLS reduction as literal elapsed time.
Assumptions & Important Notes
- Every example compares control and variant data collected with a consistent metric definition and comparable measurement conditions.
- The number of variant users represents the applicable exposure count for the conversion calculation.
- Value per conversion is an assigned average value and is not adjusted for costs or downstream outcomes.
- The examples show arithmetic only and do not establish statistical reliability or causation.
Related Examples
Frequently Asked Questions
Can I use these examples for LCP, INP, and CLS?
Yes, provided the two groups use comparable measurement methods. Use scaled CLS inputs consistently.
What does a zero incremental value mean?
It means the entered conversion rates are the same, so the formula estimates no conversion-based value difference.
Why can a performance improvement have negative estimated value?
The calculator uses the observed conversion difference. Other variant changes, random variation, or experiment issues may affect that result.
Should I multiply the calculation by all site traffic?
Use only the users exposed to the variant for the measured test-period estimate. Scaling beyond the test requires separate assumptions.
Ready to calculate your own result?
Use the live calculator with your own inputs, timing, and preferences.