What is a good PageSpeed Insights score?
Google PageSpeed Insights gives every page a performance score from 0 to 100, and a good PageSpeed Insights score on mobile is 90 or more. It is the number customers, competitors and web agencies quote most, and the one most often misread. This guide explains what the score means, what sits behind it and what usually holds SME websites back.
In Google PageSpeed Insights, a performance score of 90–100 is good, 50–89 needs improvement and 0–49 is poor. The score comes from a lab test on an emulated mid-range phone, so mobile scores are usually lower than desktop scores and vary from run to run. Test three times on mobile and use the middle result.
Which PageSpeed Insights score bands count as good?
A score of 90 or more is good, 50–89 needs improvement and anything below 50 is poor. PageSpeed Insights shows the bands in colour (Chrome for Developers):
| Score | Colour | Rating |
|---|---|---|
| 90–100 | Green | Good |
| 50–89 | Orange | Needs improvement |
| 0–49 | Red | Poor |
A score of 100 is not necessary. A page that scores in the 90s on mobile is fast for real visitors, and the last few points rarely change anything a customer notices.
What does the PageSpeed score measure?
It measures how quickly a page shows its content, how soon it responds and how much it jumps around while loading. The score combines five lab measurements, each with its own weight (Chrome for Developers):
| Metric | What it means in plain words | Weight |
|---|---|---|
| Total Blocking Time (TBT) | How long the page is too busy running scripts to respond to a tap | 30% |
| Largest Contentful Paint (LCP) | When the biggest element, usually the main image or headline, appears | 25% |
| Cumulative Layout Shift (CLS) | How much the layout jumps while the page loads | 25% |
| First Contentful Paint (FCP) | When the first text or image appears | 10% |
| Speed Index | How quickly the visible part of the page fills in | 10% |
- Largest Contentful Paint
- 2.0 s
- Total Blocking Time
- 0 ms
- Cumulative Layout Shift
- 0
rakennusalaoja.fi, Google PageSpeed Insights, mobile, 28 Oct 2026. Linked run shown; the median of three runs that day was 98. Open the report
A sixth metric, Interaction to Next Paint (INP), measures how quickly a page reacts when real people tap or click. A lab test has no real user, so it can't measure INP and uses TBT as a stand-in (web.dev, 2024). INP only appears in field data, which the next section explains.
What is the difference between lab data and field data?
Lab data is one simulated visit on a set device and connection, while field data is what real Chrome users experienced on your page. The score at the top of PageSpeed Insights is lab data from Google's Lighthouse tool. Field data comes from the Chrome User Experience Report (CrUX) and covers the previous 28 days (Google for Developers).
Field data decides whether a page passes Google's Core Web Vitals. A page passes when, for 75% of visits, all three of these are good (web.dev, 2024):
| Core Web Vital | Good | What it measures |
|---|---|---|
| LCP | 2.5 seconds or less | Loading |
| INP | 200 milliseconds or less | Responsiveness |
| CLS | 0.1 or less | Visual stability |
Across the web, 48% of mobile sites had good Core Web Vitals in Google's real-user data in July 2025, up from 36% in 2023 (HTTP Archive, 2026).
Most SME pages never get field data, because CrUX needs enough visitors to be statistically reliable. In our study of 85 SME websites, only 14 had field data for the page we tested (Visuco, 2026). The two can also disagree: the 12 pages that passed Core Web Vitals in field data had lab scores anywhere from 48 to 99. That is why the lab score is the number you can check on any site, any day.
Why are mobile scores lower than desktop scores?
Because the mobile test simulates a mid-range phone on a slow mobile connection, while the desktop test simulates a fast computer on a wired connection. In our measurements, PageSpeed Insights emulated a Moto G Power phone on a simulated slow 4G connection, using Lighthouse 13.5.0. Under those conditions even very light pages took 1.1–2.5 seconds to show their main content, while the same pages appeared in 0.3–0.8 seconds on desktop (Visuco, 2026).
Across the 85 websites we tested, 73 scored higher on desktop than on mobile. The median desktop score was 98 and the median mobile score 84 (Visuco, 2026). When someone tells you their site scores 99, ask whether that is the mobile score.
Why does my PageSpeed score change every time I test?
Because the test runs over a real network on real hardware, and the page itself may change between visits. Google lists network conditions, the testing machine's hardware and other programs competing for it, and changing page content such as ads and A/B tests as the main causes (Lighthouse documentation). Lighthouse's own documentation says the median of five runs is twice as stable as a single run.
We saw the same in our own study. We tested 13 sites three times each. The gap between a site's best and worst run ranged from 0 to 31 points, with a median of 4, and two of the 13 sites landed on both sides of 90 depending on the run (Visuco, 2026). The lightest sites barely moved: the three repeated sites with a median of 95 or more varied by 2 points at most.
The fair way to test is to run PageSpeed Insights three times on mobile and use the middle result. That is also how our speed guarantee is measured.
How fast are SME websites?
It depends heavily on who built them and how. On 28 October 2026 we tested 85 SME websites. Client sites from studios that write their own code scored far higher on mobile than client sites from template, WordPress and site-builder agencies (Visuco, 2026).
Illustration. Times are medians of Largest Contentful Paint (mobile, PageSpeed Insights) from our measurement of 85 SME websites on 28 Oct 2026: 30 client sites of hand-coding studios and 12 client sites of template and WordPress agencies. Read the study
The method, all six groups and the limits of the data are in the website speed study 2026.
What slows SME websites down?
Mostly the main image. In our study, 35 of the 85 pages took more than 4 seconds to show their largest element on mobile, while only 6 had a layout shift above 0.1 and only 6 had more than 200 milliseconds of blocking time (Visuco, 2026). Every page with a mobile LCP of 2.5 seconds or less scored 90 or more.
These are the usual causes:
- Oversized hero images and sliders. A 3 MB photo straight from the camera, or five slides that all load at once.
- Images without set dimensions. The browser doesn't know how much space to keep, so the text jumps when the image arrives.
- Page builders and plugins. Each one adds code to every page, whether that page uses it or not.
- Third-party widgets. Chat bubbles, booking widgets, review carousels and extra tracking tags, each loading its own scripts.
- Web fonts loaded late. Several font files, or fonts that block the text from appearing.
- A slow or distant server. Every request waits before anything else can start.
On one site we built, PageSpeed Insights flagged 892 KiB of possible savings in image delivery and images without fixed width and height. Its mobile score was 75, 79 and 81 over three runs, while another site we built, with optimised images, scored 99, 98 and 98 (Visuco, 2026).
Does PageSpeed affect Google rankings?
Yes, but relevance comes first. Google says Core Web Vitals are used by its ranking systems, but also that Search always tries to show the most relevant content, even if its page experience is poor (Google Search Central, 2026). A fast page about the wrong thing will not outrank a slower page that answers the question.
Speed matters most for the people who arrive, whether they came from Google, an ad or an AI assistant. If AI search is your focus, see how to get recommended by ChatGPT. If you pay for ads, the landing page structure that sells shows what the fast page should say.
Questions about PageSpeed scores
Do I need a score of 100?
No. Anything from 90 to 100 is rated good, and a page in the 90s on mobile is fast for real visitors. Chasing the last few points rarely changes anything a customer notices.
My site scores 95 on desktop and 60 on mobile. Which one counts?
Mobile. It is the harder test and closer to how many customers see your site. In our study, 73 of 85 sites scored higher on desktop than on mobile.
Why does PageSpeed Insights say there is no data from real users?
Field data from CrUX needs enough visitors to be statistically reliable, and most SME pages don't get that many. Only 14 of the 85 pages in our study had it. Use the lab score instead.
Can a WordPress site score 90 or more?
It can, but it takes careful work. In our lab tests, none of the 8 WordPress pages reached 90 on mobile and the best scored 74 (Visuco, 2026). In real-user data, 45% of WordPress sites passed Core Web Vitals on mobile in 2025, against 74% on Wix and 85% on Duda (HTTP Archive, 2026).
How do I test my site properly?
Go to pagespeed.web.dev, enter the page address, choose the mobile result and run the test three times. Use the middle score, and test the pages that matter most, such as your home page and your main service page.
Is Google PageSpeed Insights accurate?
The lab score varies between runs because it simulates a mid-range phone on a slow connection, so take the median of three runs. The field data, when your site has enough visitors, shows what real users experienced.
How do you optimize website speed?
Serve small images in modern formats, load fonts and scripts only where needed, keep the HTML plain, and host close to your visitors. Hand-coded sites do this by design; page builders usually need plugins to get close.
Sources
- Chrome for Developers, "Lighthouse performance scoring", accessed October 2026 developer.chrome.com/docs/lighthouse/performance/performance-scoring
- web.dev, "Web Vitals", October 2024 web.dev/articles/vitals
- Google for Developers, "About PageSpeed Insights", accessed October 2026 developers.google.com/speed/docs/insights/v5/about
- HTTP Archive, "Web Almanac 2025: Performance", January 2026 almanac.httparchive.org/en/2025/performance
- Visuco, "SME website speed study 2026", October 2026 /research/website-speed-study-2026
- GoogleChrome/lighthouse, "Score variability", accessed October 2026 github.com/GoogleChrome/lighthouse/blob/main/docs/variability.md
- Google Search Central, "Understanding page experience in Google Search results", September 2026 developers.google.com/search/docs/appearance/page-experience
- HTTP Archive, "Web Almanac 2025: CMS", January 2026 almanac.httparchive.org/en/2025/cms
Keep reading
How to get your business recommended by ChatGPT and other AI assistants
How AI assistants find and choose businesses, and the steps an SME can take on its own website and elsewhere to become one they mention.
Read the guideGoogle AI Overviews and AI Mode: what they mean for local businesses
What Google's AI Overviews and AI Mode change for local businesses, with click and prevalence data, and what still brings in calls.
Read the guideThe landing page structure that sells: 11 parts in the right order
The eleven parts of a landing page that sells, why the order matters and how the structure looks for a local service business.
Read the guide