Paste your web address into Google’s PageSpeed Insights, press Analyze, and a few seconds later you get a page full of coloured scores, unfamiliar abbreviations and a long list of suggestions. Most people look at the big number in the circle, feel either pleased or worried, and close the tab.
That is a missed opportunity. The number in the circle is the least useful part of the report. The most useful part sits above it, and it tells you how real visitors experience your website. This guide explains what each section means, which figures are worth acting on and where to start when the report is full of red.
What PageSpeed Insights actually checks
PageSpeed Insights, often shortened to PSI, is a free Google tool. It reports on a single page, for both mobile and desktop, and offers suggestions for improving it. It does this by combining two different types of data, and mixing them up is the most common reason people misread their results.
- Field data comes from real people. It is drawn from the Chrome User Experience Report, which collects anonymised speed measurements from Chrome users who have opted in. PSI shows what those visitors experienced over the previous 28 days.
- Lab data comes from a test. PSI runs Lighthouse, which loads your page once in a controlled setting and records what happens. For mobile, it simulates a mid-range phone on a slower mobile connection. For desktop, it simulates a desktop computer on a wired connection.
Field data tells you what is happening to your visitors. Lab data helps you find out why. You need both, but if the two disagree, trust the field data.
Start with the field data section
At the top of the report, look for the heading that says whether the page has passed the Core Web Vitals assessment. This is the part Google cares about most, and it rests on three measurements:
- Largest Contentful Paint (LCP): how long the biggest visible element, usually a main image or headline, takes to appear.
- Interaction to Next Paint (INP): how quickly the page responds after someone taps, clicks or types.
- Cumulative Layout Shift (CLS): how much the page jumps around while it loads.
Each one is placed into a band of good, needs improvement or poor, using these thresholds:
| Measurement | Good | Needs improvement | Poor |
| LCP | Up to 2.5 seconds | 2.5 to 4 seconds | Over 4 seconds |
| INP | Up to 200 milliseconds | 200 to 500 milliseconds | Over 500 milliseconds |
| CLS | Up to 0.1 | 0.1 to 0.25 | Over 0.25 |
The figure shown for each measurement is the 75th percentile. Put simply, three out of four visits were as fast as that number or faster. Google chose this on purpose, so that a page is judged on how it treats most of its visitors, including those on older phones and poor connections, rather than on its best-case visit.
A page passes only when all three measurements are good at that 75th percentile.
What if there is no field data?
Smaller sites often see a message saying there is not enough data. Chrome only reports on pages with enough visits, and the page must be public and indexable. If the individual page falls short, PSI may fall back to figures for the whole website. If the website falls short too, you will see nothing in this section.
That is not a fault. It simply means the lab data is all you have for now, so treat it as a rough guide rather than a verdict.
Then look at the lab scores
Beneath the field data sit four dials: Performance, Accessibility, Best Practices and SEO. A score of 90 or above counts as good, 50 to 89 needs improvement, and anything under 50 is poor.
Two points are worth knowing before you react to these.
- The score will change between tests. Running the same page twice can give different results, because of network conditions and the load on the testing hardware. Run it three times and take the middle result rather than chasing a single high one.
- A green dial does not guarantee happy visitors. The test uses one simulated device. Your real audience uses hundreds of different ones.
The Performance dial also lists further metrics, including First Contentful Paint, Total Blocking Time and Speed Index. They are useful for diagnosis, but LCP, INP and CLS are the ones that make up the Core Web Vitals assessment, so they deserve your attention first.
Turning the report into a to-do list
Further down, PSI lists opportunities and diagnostics. Do not try to clear every line. Work out which Core Web Vitals measurement is failing, then match it to its usual cause.
If LCP is slow, find the element PSI names as the largest on the page. It is very often a hero image that is too large, not compressed, or loaded late. Resize it, convert it to a modern format such as WebP, and make sure it is not held back by scripts that load before it.
If INP is poor, the cause is nearly always too much JavaScript running at once. Chat widgets, tracking tags, sliders and booking plug-ins are common culprits. Remove anything that is not earning its place, and delay the rest until the page has finished loading.
If CLS is high, something is pushing content around as the page loads. Images and videos without set dimensions, cookie banners that appear late and adverts that load into empty space are the usual causes. Reserve the space for them in advance.
If the server is slow to respond, the experimental Time to First Byte figure will show it. Good hosting and proper caching make a bigger difference here than any page-level tweak. It should ideally come in under 800 milliseconds.
On a WordPress website, much of this comes down to the theme, the plug-ins and the hosting. A well-built theme with a lean set of plug-ins is the cheapest speed improvement available, and it is why WordPress web design and optimisation is worth getting right from the start. Once the work is done, ongoing website support keeps updates and new plug-ins from quietly undoing it.
Does a better score mean better rankings?
Core Web Vitals form part of Google’s page experience signals, so they matter. They are not a shortcut, though. A fast page with weak content will not beat a relevant, useful page that is a little slower. Speed works best as a tiebreaker and as a way of keeping visitors who have already found you.
That second point is where the money is. A visitor who waits for a slow page, or who taps a button that does nothing, leaves. A page can rank well and still lose enquiries for exactly that reason, which is the same problem covered in why your website can rank number one and still fail you.
A sensible routine
- Test your most important pages, not just the homepage. Service pages, product pages and landing pages are where visitors make decisions.
- Check mobile first. Most visits come from phones, and that is where problems show up.
- Fix one measurement at a time, then retest after a few days.
- Allow 28 days for the field data to catch up with your changes.
- Keep a record of results, ideally alongside your other reporting and analytics, so you can see whether speed improvements line up with fewer exits and more enquiries.
PageSpeed Insights is free, and it gives you a clear view of what Google can measure about your pages. If the report points to problems you would rather not tackle yourself, our SEO services include technical fixes, or you can get in touch for a free audit of your website.
Graig Upton
Graig Upton is a UK SEO and Google Ads consultant with 23 years of hands-on digital marketing experience. Google certified in Ads, Analytics and Conversions, he has helped businesses ranging from local providers to national brands - including Nando's and Financial Times - dominate search and scale their leads. Based in Preston, working with clients across the UK.