What your Google PageSpeed score actually means
Learn what the Performance score measures, how to read real-visitor data, and what to check first.
Short answer: The 0–100 Performance score comes from Lighthouse testing one URL under simulated mobile or desktop conditions. It combines measures of loading, visual stability, and time the browser spends blocked. Google rates 90–100 as good, 50–89 as needing improvement, and 0–49 as poor.
PageSpeed Insights is Google's free tool for checking how a page performs. You put in a URL, and it hands back a performance score from 0 to 100, plus a list of specific things slowing the page down. You'll actually see four scores: Performance, Accessibility, Best Practices and SEO. This article is about Performance, the one people usually mean. The "SEO" score is a basic checklist (page titles, descriptions, links), not a prediction of where you rank.
The report contains two different kinds of data
This is the part most people miss, and it explains most of the confusion.
- Lab data. Google loads your page in a controlled test environment, on a simulated mid-range phone over a mobile connection (or a desktop on a wired connection, if you pick Desktop). That is why the mobile score is often noticeably lower than the desktop one for the same page. This is where the 0 to 100 performance score comes from. Because it is a simulation, the number can move a few points between runs on an unchanged page — it is a diagnostic, not a certificate.
- Field data. Measurements from real Chrome users who actually visited your page. Check whether the report shows data for the tested URL or the whole site. Google can show site-wide data when there isn’t enough for the individual page. If neither has enough data, this part of the report is unavailable.
Both are described in Google's own documentation: About PageSpeed Insights, developers.google.com. Worth reading if you want the primary source rather than someone's summary of it — including ours.
"Passed" or "Failed": the Core Web Vitals line
If Google has enough real-visitor data for your site, the top of the report shows a Core Web Vitals assessment instead of a score. It is based on three measurements from real Chrome users over the previous 28 days:
- Largest Contentful Paint (LCP): how long the main content takes to appear. Good is 2.5 seconds or less.
- Interaction to Next Paint (INP): how quickly the page responds when someone taps or clicks. Good is 200 milliseconds or less.
- Cumulative Layout Shift (CLS): how much content shifts unexpectedly during a visit. Good is 0.1 or less.
When all three metrics have enough data, each must meet its good threshold at the 75th percentile for the assessment to pass. For example, an LCP at or below 2.5 seconds means at least three quarters of measured visits met that loading threshold.
In our September 23, 2026 snapshot of six agency homepages, five scored between 50 and 89 in a single mobile test. The sample shows how far results can vary across pages. It covers those six homepages only.
Two things the score is not
It is not a revenue figure. You will see people convert a score into lost customers or lost dollars. Nothing in Google's documentation supports that arithmetic, and we are not going to invent it. Speed matters to people trying to use your site — that is a reasonable thing to believe on its own, without a fabricated number attached to it.
It is not your Google ranking. Google says Core Web Vitals, the real-visitor measurements above, are used by its ranking systems. The 0–100 lab score is not. Google also says it "always seeks to show the most relevant content, even if the page experience is sub-par" (Understanding page experience, developers.google.com). A site can score 95 and rank poorly because it has no content anyone searched for, and a site can score 60 and rank first because it is the most relevant result. Speed is one factor among many.
What to actually do with the report
- Enter a page that customers land on, such as a key service page, in PageSpeed Insights.
- Start with Mobile and check Desktop too. Rerun the same URL in the same mode before acting on a small score change.
- Choose a finding you can connect to a visitor’s experience. If the main image appears slowly and the report identifies it as a bottleneck, investigate that image first.
- After a change, rerun the same test and compare the specific finding and metric you worked on.