PageSpeed Insights is Google’s free tool that measures a page’s speed and shows what slows down loading. The report has two sections, real user data and a lab simulation, and the 0 to 100 score comes only from the second. To fix issues, start with LCP on mobile, which is the time until the largest element on the screen appears.
The first time you open the tool, you see a red or orange circle, a long list of warnings and no order of priority. The common reaction is to try to clear every warning. It’s the worst strategy, because half of them change very little about what the visitor actually feels.
This guide shows you how to read the screen, what each number means and the fix order that usually works. It also includes what we measured on real company sites, so you can compare with yours.
How PageSpeed Insights works
The tool combines two types of data: field data, which comes from real Chrome users, and lab data, which comes from a simulation. Google explains this in the official PageSpeed Insights documentation.
Field data comes from the Chrome User Experience Report and covers the previous 28 days, so it reflects real phones and real connections. Lab data runs Lighthouse in a controlled environment. In the mobile test, the simulation uses a mid-range device and a mobile network. On desktop, you get an emulated desktop on a wired connection.
That’s why the two rarely match. The documentation says field data reflects many devices and networks, while lab data reflects a single setup, and that a good lab score doesn’t guarantee a good real-world experience.
What each section on the screen means
- Discover what your real users are experiencing is the field data. It only appears if the site has enough traffic.
- Diagnose performance issues is the lab data, with the score and metrics.
- Opportunities and diagnostics is the list of suggestions, which helps you find the cause.
A new site or one with few visits has no field data. In that case, only lab data appears, and it works as a hint, not a verdict.
How to read your PageSpeed Insights score
The performance score ranges from 0 to 100 and is split into three bands: 90 to 100 (good), 50 to 89 (needs improvement) and 0 to 49 (poor). These bands are in Google’s documentation, and the score summarizes six lab metrics.
The metrics that make up performance are FCP, LCP, Speed Index, CLS, Time to Interactive and Total Blocking Time. The documentation we reviewed doesn’t list the weight of each one in the score, so we won’t state a percentage.
Three ideas help you avoid being misled.
- The score changes from one test to the next on the same page, because the simulation has small variations. Run it three times and look at the range.
- A high score is not the goal. The goal is for the page to load fast for people coming from your ad or from Google.
- Mobile and desktop have different scores. Prioritize mobile, which is where most of a local business’s traffic comes from.
The metrics worth your attention
Of the metrics in the report, three make up the Core Web Vitals and are the ones Google uses to rate page experience. They are LCP, INP and CLS, and the full explanation is in our Core Web Vitals guide.
| Metric | What it measures | Limit considered good |
|---|---|---|
| LCP | Time until the largest element on the screen appears | up to 2.5 seconds |
| INP | How fast the page responds to taps and clicks | up to 200 milliseconds |
| CLS | How much the layout shifts while the page loads | up to 0.1 |
The thresholds come from the Core Web Vitals page on web.dev, measured at the 75th percentile of visits, with mobile and desktop reported separately. A page passes when all three metrics meet the target at that percentile. In the report, if INP doesn’t have enough data, the assessment considers only LCP and CLS.
What we measured on business websites
On real business websites, mobile LCP is often above 2.5 seconds, the limit Google considers good. In October 2026, we pulled the profiles Google Maps shows first in São Paulo, Rio de Janeiro, and Belo Horizonte for 15 niches, and ran a sample of 8 websites per niche through the test on mobile. For real estate agencies, the sample was 7.
These are the numbers for five niches, each with 8 websites measured.
| Niche | Mobile LCP (median) | Websites with LCP above 2.5 s | Performance score (median) |
|---|---|---|---|
| Law firms | 4.95 s | 88% (7 of 8) | 59 |
| Dentists | 5.4 s | 100% (8 of 8) | 72 |
| Aesthetics | 5.05 s | 100% (8 of 8) | 62,5 |
| Accounting | 5.15 s | 100% (8 of 8) | 62,5 |
| Architecture | 15.45 s | 100% (8 of 8) | 55,5 |
Two things to keep in mind when reading this. The sample is small, 8 sites per niche, and comes from São Paulo, Rio, and Belo Horizonte, so it is not a picture of Brazil or of the market as a whole. And the architecture median LCP of 15.45 seconds shows what happens when the first element on the page is a huge, unoptimized photo.
Use the data as a yardstick. If your site opens on mobile in 3 seconds, you are ahead of a good share of the businesses showing up on Google Maps in your niche. If it opens in 8, you are in the typical range, and the typical range is far from the target.
What to fix first
Fix these in this order, from what weighs most on LCP to what weighs least, and measure again after each step. Changing everything at once hides which fix actually worked.
- The main image above the fold. On most slow sites, the largest element is a photo or a banner. Reduce its file size, use a modern format, set width and height, and don’t lazy-load this image.
- Server response time. If your hosting is slow to deliver the HTML, nothing after that makes up for it. Page caching solves a good share of cases on WordPress.
- Third-party scripts. Chat, pixel, map, embedded video and tags piled up in Google Tag Manager slow the page down. Remove what nobody uses and defer the rest.
- Render-blocking CSS and JavaScript. Page builder plugins load a lot that the page doesn’t use.
- Fonts. Few families, few weights, loaded without hiding the text.
- Layout shift. Reserve space for images, ads and the cookie banner so the content doesn’t jump.
If your site runs on WordPress, the process has plugin and theme specifics, which we explain in SEO for WordPress. For the bigger picture, and where speed fits among the checks that come before producing content, see the technical SEO guide.
How much people search for this
In October 2026, “pagespeed insights” gets about 6,600 searches per month in Brazil and “google pagespeed insights” gets 2,900, according to Google Keyword Planner. The estimated bid for the first one, at the top of the page, is R$5.05.
Whoever types this is almost always someone holding a red report and unsure what to do. The interest exists because the report is intimidating.
How Ciclicko uses the report
We use the report as a starting point, never as the deliverable. We measure the page on mobile, cross-check it with field data when available, and turn the warnings into a short list of fixes, ordered by impact on LCP.
After each change, we measure again and record the before and after. The report becomes part of our website optimization work alongside content and structured data, because speed alone won’t put anyone on the first page.
Frequently asked questions
What is PageSpeed Insights? It’s Google’s free tool that measures the speed and experience of a page, on mobile and desktop, and suggests fixes. It shows real-user data when available and a lab simulation.
What is a good score? According to Google’s documentation, 90 to 100 is considered good, 50 to 89 needs improvement, and below 50 is poor. The score comes from a lab test and varies between runs.
Why does my score change with every test? Because the lab simulates a device and a network, and small variations in the server and the connection change the result. Run three tests and use the range.
What’s the difference between field data and lab data? Field data comes from the Chrome User Experience Report and covers the previous 28 days. Lab data is a simulation run on the spot with Lighthouse.
Does the score affect my Google ranking? What Google treats as a page experience signal is the Core Web Vitals, LCP, INP and CLS, not the 0 to 100 score. The score is a diagnostic tool.
What should I fix first? LCP on mobile, starting with the main image on the first screen and the server response time.
Want to know what’s slowing down your site?
We measure your page on mobile, separate what really weighs it down from what’s just noise, and send back a list of fixes in order. Learn about Ciclicko’s website optimization and request the analysis.