A/B Testing Core Web Vitals Monthly Impact Formula
Learn how to estimate the monthly change in good Core Web Vitals visits when an A/B test variation is rolled out to all eligible sessions.
This calculator translates the difference between control and variation good Core Web Vitals rates into an estimated monthly number of additional good visits. It also calculates the traffic-weighted good rate during the current test split, helping separate the observed test mix from the full-rollout estimate.
- 100% Free
- No Sign-Up Required
- Private & Secure
- Mobile Friendly
Monthly Additional Good Core Web Vitals Visits
Where:
Find the percentage-point difference between the variation and control good CWV rates, then apply that difference to all eligible monthly sessions.
Variables Explained
| Variable | What It Means | Unit |
|---|---|---|
| monthlySessions - Monthly eligible sessions | The estimated number of monthly sessions that could receive either the control or variation experience. | sessions |
| controlGoodRate - Control good CWV rate | The percentage of control sessions classified as having a good Core Web Vitals experience. | percent |
| variationGoodRate - Variation good CWV rate | The percentage of variation sessions classified as having a good Core Web Vitals experience. | percent |
| variationTrafficPercent - Variation traffic allocation | The percentage of eligible sessions assigned to the variation during the test. | percent |
| controlTrafficShare - Control traffic share | The decimal share of eligible sessions assigned to the control group. | / |
| variationTrafficShare - Variation traffic share | The decimal share of eligible sessions assigned to the variation group. | / |
Step-by-Step Calculation
Convert the traffic allocation to shares
The selected allocation determines how eligible sessions are split between the two test experiences.
controlTrafficShare = (100 - variationTrafficPercent) / 100; variationTrafficShare = variationTrafficPercent / 100
Estimate sessions in each group
Multiply eligible monthly sessions by each traffic share.
controlSessions = monthlySessions * controlTrafficShare; variationSessions = monthlySessions * variationTrafficShare
Estimate good CWV visits during the test
Apply each group's observed good CWV rate to its estimated sessions.
controlGoodVisits = controlSessions * (controlGoodRate / 100); variationGoodVisits = variationSessions * (variationGoodRate / 100)
Calculate the blended test rate
The blended rate is a traffic-weighted result, so it changes when the traffic allocation changes.
combinedGoodRate = ((controlGoodVisits + variationGoodVisits) / monthlySessions) * 100
Project both experiences at full traffic
For a rollout comparison, apply each rate to the entire eligible monthly audience.
fullRolloutControlGoodVisits = monthlySessions * (controlGoodRate / 100); fullRolloutVariationGoodVisits = monthlySessions * (variationGoodRate / 100)
Calculate the monthly rollout difference
A positive result means the variation is estimated to create more monthly good CWV visits than the control.
monthlyGoodVisitDifference = fullRolloutVariationGoodVisits - fullRolloutControlGoodVisits
Example: 100,000 eligible sessions with a 50/50 test split
Split test traffic
100,000 × 50% for each group
50,000 control sessions and 50,000 variation sessions
Estimate control good visits
50,000 × 65%
32,500 good CWV visits
Estimate variation good visits
50,000 × 72%
36,000 good CWV visits
Calculate blended test rate
(32,500 + 36,000) ÷ 100,000 × 100
68.5%
Calculate full-rollout difference
100,000 × (72% − 65%)
7,000 additional good CWV visits per month
Final Result
If the variation's 72% good CWV rate remains representative at full traffic, it is estimated to produce 7,000 additional monthly visits with a good Core Web Vitals experience versus the 65% control rate.
Assumptions
- ✓Control and variation rates use the same good Core Web Vitals definition and measurement method.
- ✓The observed test groups represent the audience that would receive a full rollout.
- ✓Eligible monthly session volume remains broadly similar after rollout.
- ✓The entered rates are treated as stable monthly estimates.
Limitations
- !The calculation does not test statistical significance, confidence intervals, or sample adequacy.
- !Changes in traffic mix, devices, locations, page types, or seasonality can change rollout results.
- !A higher good CWV rate does not by itself establish changes in conversion, revenue, rankings, or other outcomes.
- !Sessions may not be independent observations if the same people visit repeatedly.
Common Mistakes to Avoid
Using all site traffic instead of only sessions genuinely eligible for either experiment experience.
Comparing rates measured over different time periods, device groups, countries, or page templates.
Reading a 65% to 72% change as a 72% increase rather than a 7 percentage-point increase.
Treating the blended rate from an uneven traffic split as the full-rollout rate.
Assuming a calculated uplift proves that the variation caused the observed difference.
Related Formulas
Frequently Asked Questions
How do you calculate additional good Core Web Vitals visits?
Multiply eligible monthly sessions by the difference between variation and control good CWV rates expressed as a decimal. For example, 100,000 sessions times a 7 percentage-point difference equals 7,000 visits.
Why is the rate change shown in percentage points?
Percentage points show the direct gap between two rates. A move from 65% to 72% is 7 percentage points, not a 7% relative increase.
Does traffic allocation affect the full-rollout estimate?
No. The full-rollout comparison applies each entered rate to all eligible sessions. Allocation affects the estimated blended rate during the active test.
What does a negative monthly result mean?
It means the variation's entered good CWV rate is below the control rate, so rolling out the variation would be estimated to reduce monthly good CWV visits at the stated traffic level.
Does this formula establish statistical significance?
No. It estimates practical monthly volume from the rates entered. Statistical testing requires separate analysis of sample size, uncertainty, duration, and potential confounding factors.
Ready to calculate your result?
Use the calculator to get instant results with your own inputs.