Comparisons
Core Web Vitals vs Lighthouse Score: Which One Google Uses
One is a measurement of your real users. The other is a weighted average of simulated metrics. Only the first affects ranking, and most teams optimise the second.
Run a fresh DomainLens audit and use the report as your priority list.
They are not the same measurement
Core Web Vitals are three specific metrics — LCP, INP and CLS — collected from real Chrome users and aggregated over 28 days. The Lighthouse performance score is a single 0–100 number computed from a weighted blend of lab metrics measured during one simulated page load.
A site can pass Core Web Vitals comfortably while scoring 65 in Lighthouse, and score 100 in Lighthouse while failing in the field. Neither outcome is a contradiction: they measure different populations under different conditions.
| Core Web Vitals | Lighthouse score | |
|---|---|---|
| Source | Real Chrome users | One simulated load |
| Device | Whatever visitors use | A throttled mid-range phone |
| Window | Rolling 28 days | Right now |
| Output | Three metrics, pass or fail | One score out of 100 |
| Used for ranking | Yes | No |
| Best for | Knowing if there is a problem | Finding what causes it |
Only the field metrics affect ranking
Google has been explicit and consistent: page experience uses Core Web Vitals from the Chrome UX Report. The Lighthouse performance score is a diagnostic tool and is not a ranking input at all.
This matters because the score is the number everyone screenshots. Reporting "we improved the score from 72 to 94" describes an improvement in a simulation, and it may correspond to no change whatsoever in what real users experience.
Why the score can mislead
The INP gap is the most important. A page can score 100 in Lighthouse and be unusable on interaction, because the score literally cannot see the metric that would reveal it.
- It is a weighted blend, so a large gain in a lightly weighted metric moves the number without helping anyone.
- It varies between runs on the same page, sometimes by ten points, because it is one sample.
- It cannot measure INP at all, because a lab run never interacts with the page.
- It simulates one device profile, which may be far better or worse than your actual audience.
- It rewards fixes that are invisible to users, like shaving bytes off an already-cached asset.
Using both correctly
The order matters: field data identifies the problem, lab data explains it. Reversing that produces a lot of optimisation on pages that were never failing. See Search Console vs PageSpeed Insights for where each lives, and PageSpeed Insights vs Lighthouse vs CrUX for the three data sources.
- 1Start with field data in Search Console, which groups URLs and tells you which templates fail.
- 2Take one failing URL into Lighthouse to find the cause — the LCP element, blocking scripts, long tasks.
- 3Fix the cause at the template level.
- 4Confirm in the lab that the specific diagnostic improved, not that the score went up.
- 5Wait for field data to confirm. The 28-day window means real verification takes weeks.
How DomainLens contributes
DomainLens reports the field metrics where they exist and the lab diagnostics where they do not, labelled as what they are — so a report never presents a simulated score as though Google reads it. See the Core Web Vitals guide for the metrics themselves.
- Does the Lighthouse score affect rankings?
- No. Google uses Core Web Vitals from real users. The Lighthouse performance score is a diagnostic and is not a ranking input.
- Why does my score change between runs?
- It is a single simulated load, subject to network variance and CPU contention. Variation of several points on the same page is normal.
- Can I pass Core Web Vitals with a low Lighthouse score?
- Yes, routinely. The score blends metrics that field data does not use, and it is measured on a device profile that may not match your visitors.
- Is a 100 score worth pursuing?
- Rarely. Once field metrics pass, further lab-score work optimises a simulation. The same hours spent on content or intent usually return more.