Short answer: PageSpeed Insights tests how a page loads on phones and computers. Its report has two parts. Field data shows real Chrome users over the last 28 days, when there’s enough of it. A lab test by Lighthouse gives the 0 to 100 score. A score of 90 or above is good, and the place to start is the list of opportunities under the score.
The weight of each metric in the score
| Lab metric | What it measures | Weight in the score |
|---|---|---|
| Total Blocking Time | How long the page is too busy to respond | 30% |
| Largest Contentful Paint | When the largest content appears | 25% |
| Cumulative Layout Shift | How much the layout jumps | 25% |
| First Contentful Paint | When the first content appears | 10% |
| Speed Index | How fast the content fills the screen | 10% |
The score is a weighted average of these five lab metrics, as set in Lighthouse 10. Only the metrics count: the opportunities and diagnostics don’t add points by themselves, but fixing them usually improves the metrics.
From Chrome for Developers, “Lighthouse performance scoring”, read November 3, 2026.
Field data and lab data
Field data comes from the Chrome User Experience Report and covers real visits over the previous 28 days. PageSpeed Insights shows the 75th percentile of each metric, so you see the more frustrating visits, not the average.
A page passes the Core Web Vitals assessment when LCP, INP and CLS are all good at that percentile, or LCP and CLS when INP lacks data. If the page lacks data, the report falls back to the whole site, and with too little data it shows none.
Lab data comes from a single simulated load, a mid-tier phone on a mobile network or an emulated computer on a wired connection. It helps you find problems, but it may miss what slows real visitors down. Our guide to Core Web Vitals covers the three field metrics.
How to read your report
- Go to pagespeed.web.dev and enter the address of one page, not only your home page.
- Look at the field data first, if the page has any, and check the Core Web Vitals assessment.
- Switch between the mobile and desktop tabs, since each has its own results.
- Read the lab metrics and note which ones are amber or red.
- Open the opportunities with the largest estimated savings.
- Fix one thing at a time, starting with what delays the main content.
- Run the test a few times after each change and compare the trend.
Images are a common cause of a slow LCP; our guide to lazy loading images covers which ones to defer.
What it shows for one of our guides
At 1:45 AM on November 3 we ran our robots.txt guide through PageSpeed Insights. The Chrome User Experience Report had no data for the page or for our site yet, so the report showed lab results only.
On mobile the page scored 83, with an LCP of 2.1 seconds, no layout shift and 400 milliseconds of total blocking time. On desktop it scored 100. The two mobile opportunities were to reduce unused JavaScript, about 270 milliseconds, and to minify JavaScript, about 120.
On that page, both opportunities are about JavaScript. Total blocking time, the lab metric with the most weight in the score, was 400 milliseconds.
Before you trust a score
- You tested the page you care about, not only the home page.
- You read the field data and the Core Web Vitals assessment first.
- You checked mobile and desktop separately.
- You know which lab metric weighs the most in your score.
- You ran the test more than once.
- You fixed the opportunities with the largest savings first.
RememberReal-user data shows what visitors experience, and the lab score helps you find the causes. Fix the opportunities that save the most time, and judge the result over several runs, not one.
If you want us to make your pages faster, see our web design service.
Quick answers about PageSpeed Insights
Why is my mobile score lower than my desktop score?
Lighthouse tests mobile as a mid-tier phone, a Moto G4, on a mobile network, and desktop as an emulated computer on a wired connection. Desktop also uses its own scoring curves, so the two scores aren’t meant to match.
Which PageSpeed Insights score counts as good?
A score of 90 or above is good, 50 to 89 needs improvement and below 50 is poor. Google adds that good lab data doesn’t necessarily mean real users have a good experience too.
Why doesn’t my page show real-user data?
The Chrome User Experience Report needs a public page and enough distinct samples. A new page or one with few visits falls back to data for the whole origin, and if the origin is short of data too, the report shows none.
Why does the score change when I run it again?
Google says measurements vary with local network availability, client hardware availability and client resource contention. Run the test a few times and compare the trend rather than a single number.
Do I need a score of 100?
No. Chrome’s documentation calls a perfect 100 extremely challenging and not expected. Going from 99 to 100 takes about the same improvement as going from 90 to 94.
Sources
Read on November 3, 2026. Google: About PageSpeed Insights. Chrome for Developers: Lighthouse performance scoring. Our own case: PageSpeed Insights for our robots.txt guide, mobile and desktop, on November 3, 2026.