Skip to content
arduralab
·8 min

PageSpeed Insights: Field Data vs. the Lighthouse Score

Core Web Vitalstechnical SEOLighthouseSEO tools
MG
Marcin Godula

Współzałożyciel & Head of SEO/Tech

Specjalista SEO, GEO i web development z ponad 15-letnim doświadczeniem. Pomaga firmom B2B budować widoczność w wyszukiwarkach klasycznych i AI.

PageSpeed Insights shows two different things for a single URL. The top section is field data from the Chrome User Experience Report (CrUX): how real users experienced the page over the previous 28 days, with a Core Web Vitals assessment based on the 75th percentile. The bottom section is a Lighthouse lab test with a score from 0 to 100, meaning one simulated page load. Field data tells you whether the page is fast for people. The Lighthouse score helps you find the cause of a problem, and Google itself says good lab data doesn't necessarily mean real-user experiences will also be good.

What Is PageSpeed Insights and What Data Does It Show

PageSpeed Insights is a free Google tool that reports on page performance on mobile and desktop. According to the PageSpeed Insights documentation, the report combines two sources. Lab data is collected in a controlled environment and is useful for debugging, but it may not capture real-world bottlenecks. Field data captures true, real-world user experience, but with a more limited set of metrics.

That distinction is the key to the whole report. Both sections use similar metric names and similar colors, so it's easy to assume they describe the same thing. They don't. The field section describes what happened for people who actually visited the page. The lab section describes what happened in one test that Google ran on a chosen device and connection. When the two disagree, the tool isn't broken; they're measuring different things.

Field Data: CrUX, 28 Days and the 75th Percentile

Field data in PageSpeed Insights comes from the Chrome User Experience Report. It covers the previous 28 days for four metrics: First Contentful Paint (FCP), Largest Contentful Paint (LCP), Interaction to Next Paint (INP) and Cumulative Layout Shift (CLS), plus the experimental Time to First Byte (TTFB). The documentation adds that field data in PageSpeed Insights is updated daily.

Each metric shows a distribution across three buckets, good, needs improvement and poor, drawn as green, amber and red bars. Above the bars, the tool reports the 75th percentile. Google chose that point so pages work well for the majority of users, including under more difficult device and network conditions. The good thresholds are LCP up to 2.5 s, INP up to 200 ms and CLS up to 0.1.

The Core Web Vitals assessment follows a simple rule: a page passes when the 75th percentile of all three metrics, LCP, INP and CLS, is good. If INP has insufficient data, good LCP and CLS are enough. Without data for LCP or CLS, the page or origin can't be assessed.

Why You Sometimes See Origin Data or No Data at All

PageSpeed Insights shows field data for a URL only when CrUX has enough of it. A recently published page, or one with few visits, has no data of its own, so the tool falls back to the origin level: experiences from all pages of the site combined. Sometimes even the origin has too little data and the field section stays empty.

The CrUX methodology lists two requirements for pages and origins. The first is public discoverability, judged by the same indexability criteria search engines use: the URL must return a 200 status code after redirects and must not carry a noindex directive in a header or meta tag. The second is sufficient popularity, with a threshold that isn't disclosed. You can't submit a URL to CrUX manually.

In practice, low-traffic sites often have no field data for individual pages. The honest answer to "how fast is this page?" is then "there's no field data," not a lab score offered as a stand-in.

The 0–100 Lighthouse Score: What Goes Into It

The bottom part of the report is Lighthouse diagnostics, run in a simulated environment for the Performance, Accessibility, Best Practices and SEO categories. According to the PageSpeed Insights documentation, the mobile test simulates a mid-tier device on a mobile network, and the desktop test an emulated desktop with a wired connection. A score of 90 or above is good, 50 to 89 needs improvement, and below 50 is poor.

The Performance score is a weighted average of several metric scores. The Lighthouse performance scoring guide lists the Lighthouse 10 weights: Total Blocking Time 30%, Largest Contentful Paint 25%, Cumulative Layout Shift 25%, First Contentful Paint 10% and Speed Index 10%. Each metric is converted to a score using a distribution built from real website data in HTTP Archive. The same page stresses that a perfect 100 is extremely challenging to achieve and not expected.

Notice what's missing from that list: INP. A lab can't know when a real user will click, so it can't reliably measure responsiveness to interactions. Instead it reports Total Blocking Time, which, according to web.dev, is meant to help diagnose INP issues but doesn't consider user behavior.

Why Field Data and the Lighthouse Score Disagree

The PageSpeed Insights documentation answers this directly: field data is a historical report of real users on a variety of devices and network conditions, while lab data is a simulated load on a single device with a fixed set of network conditions, so the values may differ. The web.dev article on lab and field differences adds concrete reasons. In the field, some resources are already in the browser cache, pages may be restored from the back/forward cache (bfcache), and the LCP element depends on screen size, whether the user is logged in and personalized content.

The second cause is test variability. Google names local network availability, client hardware availability and client resource contention as common sources of fluctuation. Two runs on the same page a minute apart can produce different scores with nothing changed on the page. Field data covers 28 days, so it reacts to changes more slowly, but it doesn't jump from run to run.

For both reasons, the Lighthouse score is a poor measure of success. It's useful for comparing before and after a change under the same conditions and for the list of concrete causes in the audits section.

PageSpeed Insights vs. the Core Web Vitals Report in Search Console

Both tools draw on the same CrUX data but present it differently. Search Console Help on the Core Web Vitals report explains that the report groups similar pages into URL groups, and a group's status is the status of its worst-performing metric. PageSpeed Insights generally shows data for a single URL, so one page's numbers can differ from its group's if that page is an outlier. Search Console also distinguishes URLs by their parameters, while PageSpeed Insights strips parameters and assigns the results to the bare URL.

The division of labor follows naturally. Search Console shows which page templates across the site have a problem. PageSpeed Insights lets you inspect one URL from that group and, in the lab section, find a likely cause. In its documentation on Core Web Vitals, Google recommends achieving good results for success with Search and says that this, along with other aspects of page experience, aligns with what its core ranking systems seek to reward.

How We Read PageSpeed Insights in Audits

Field data first, then the lab. In an audit we start with the top of the report and with whether the data covers the URL or the whole origin. We treat the 0–100 score as a list of leads, not as a grade for the page. We explain it to clients the same way: we don't promise a particular number of points, because that number doesn't say how people experience the site.

Search Console before single URLs. Before testing pages one by one, we check the Core Web Vitals report to see which URL groups have a problem. In audits we regularly see the issue confined to one template, such as product pages or blog posts, rather than the whole site.

No data is a finding too. For pages without field data of their own, we record that in the report and rely on origin data. The lab test is then used to look for causes, not to hand out a grade.

INP gets its own diagnosis. Because the lab doesn't measure INP, a poor INP in field data is investigated in the browser's developer tools by reproducing real interactions on the page. It's part of a broader technical SEO audit.

Summary

PageSpeed Insights combines CrUX field data and a Lighthouse lab test in one report. Field data, covering 28 days at the 75th percentile, tells you whether the page passes the Core Web Vitals assessment and how users experience it. The 0–100 score is a weighted average from one simulated test; it changes between runs and doesn't include INP. That's why it's used for diagnosis and before-and-after comparisons, not as a target.

Sites we build as part of our web development service are assessed on field data as soon as they have traffic. If you want to find out what's slowing your site down, get a quote.

Need help with this topic?

Get a free audit and find out how we can help your business grow online.

Get a free quote