WebSpeedLab X
Home / Insights / Diagnostics

PageSpeed Insights vs Core Web Vitals: why the numbers disagree.

The Lighthouse performance score is a controlled lab result. Core Web Vitals field data describes real Chrome visits. Both are useful, but they answer different questions.

A website can score 95 in PageSpeed Insights today and still show “Core Web Vitals assessment: failed.” Another site can receive a modest lab score while its field data passes. This is not necessarily an error. The two views measure different populations, time periods and conditions.

What PageSpeed Insights contains

PageSpeed Insights can show two distinct sections. Field data comes from the Chrome UX Report when the page or origin has enough eligible traffic. Lab data is produced by Lighthouse in a simulated environment at the time of the test. The large 0–100 performance score belongs to the lab test.

The score is a weighted summary of several lab metrics. It is useful for reproducing problems, comparing controlled changes and opening diagnostics. It is not a direct record of every customer visit and should not be presented as a guaranteed search ranking.

What Core Web Vitals field data means

Field Core Web Vitals summarize real experiences over a rolling window. The assessment considers the 75th percentile and evaluates LCP, INP and CLS. Data may be available for the exact URL or only the origin, depending on traffic volume and privacy thresholds.

Search Console groups similar URLs from the same property and reports whether their field experience is good, needs improvement or poor. Because the data window rolls over time, a production fix does not instantly turn the report green.

Why lab and field results differ

CauseLabField
Device and networkOne simulated profileDistribution of real devices and networks
GeographyTest locationWhere actual visitors are
Cache stateOften a controlled first loadMix of new and returning visits
InteractionsLimited synthetic load flowReal clicks and visits measured by INP
TimeOne run nowRolling historical period
Page scopeExact tested URLURL or origin aggregation

Consent, personalization, experiments, ads and logged-in state can also produce different code paths. A test bot may not receive the same third-party workload as a real customer.

Which data should you use?

Use field data to prioritize business impact

Field data tells you whether a meaningful share of visitors experiences poor loading, responsiveness or stability. Segment real-user monitoring by device, geography, page type and release when possible. This is the right level for evaluating the real experience and following Core Web Vitals trends.

Use lab data to diagnose and verify changes

Lab tests are repeatable. They expose waterfalls, main-thread tasks, unused code and rendering timelines. Run multiple tests under the same profile, compare medians and inspect the trace instead of optimizing only the headline number.

Practical rule: field data tells you where and how broadly users suffer; lab data helps you discover why and test whether a specific change works.

Does a 100 score guarantee better rankings?

No. Search visibility depends on relevance, helpful content, links and many other systems. Page experience can support performance in search and conversions, but a perfect lab score is not a shortcut around content and authority. Chasing the last points can even damage features or measurement if scripts are disabled only to please a test.

A reliable optimization workflow

  1. Review Search Console and field data to identify affected URL groups.
  2. Select representative pages and reproduce problems in lab tools.
  3. Trace the dominant component of LCP, slow interactions or shift sources.
  4. Implement fixes in staging and run functional regression tests.
  5. Compare repeated lab runs and production telemetry.
  6. Watch the field distribution improve as new visits enter the window.

If your lab score and Search Console report disagree and the root cause is unclear, our Core Web Vitals optimization service connects the two datasets and implements the fix.

Have a green lab score but failing field data?

We will identify the affected users and templates, reproduce the issue and fix the real bottleneck.

Request a diagnostic audit