What is a speed test really telling you?
A speed test gives you two kinds of information, and only one of them reflects your customers. The big performance score at the top is a lab simulation: one test, on one throttled device, at one moment.
The Core Web Vitals figures from real visitors, shown in PageSpeed Insights and in Google Search Console, describe how the site actually behaves for people using it. When the two disagree, trust the field data.
So read the metrics, not the score. Three numbers matter most: how fast the main content appears (LCP), how quickly the page responds to taps (INP) and how much the layout jumps (CLS). Search Console groups your pages by those three and tells you which groups need work. The rest of this guide walks through each tool and report in the order you will meet them.
Lab data or field data: which should you trust?
Trust field data for the verdict and lab data for the diagnosis. Field data shows what real visitors experienced over the past weeks and feeds Google's page experience signals. Lab data is one simulated test, which makes it the right tool for finding causes and checking a fix before the field figures catch up.
Lab data
Lab data comes from Lighthouse, the engine behind PageSpeed Insights' score and inside Chrome's developer tools. It loads your page once on a simulated mid-range phone with a throttled connection. It is repeatable enough for testing changes, and it lists diagnostics. But it varies between runs, because server load, network routes and third-party scripts differ each time.
Field data
Field data comes from the Chrome User Experience Report (CrUX): anonymised measurements from real Chrome users who visited your site. It covers a rolling 28-day window and is judged at the 75th percentile, meaning three quarters of visits should meet the threshold. It moves slowly, so a fix made today takes weeks to show fully.
Which to trust
Field data shows what visitors experience and is what Google uses in its page experience signals. How much those signals, and the design choices behind them, weigh in rankings is covered in our guide to how website design affects SEO. Lab data is your workbench for finding causes and checking fixes before field data catches up. A site can score poorly in the lab and still pass in the field, and the reverse also happens.
The three Core Web Vitals
Google's thresholds below come from web.dev and apply to field data at the 75th percentile. Interaction to Next Paint (INP) replaced First Input Delay as a Core Web Vital in March 2024, so older articles that mention FID are out of date. On many business sites the LCP element is a large hero photo, which is why image optimization for faster websites is so often the first fix.
| Metric | What it measures | Good | Poor | Usual causes |
|---|---|---|---|---|
| LCP (Largest Contentful Paint) | When the main content appears | 2.5 s or less | Over 4 s | Slow server, large hero image, render-blocking CSS and fonts |
| INP (Interaction to Next Paint) | How quickly the page responds to taps and clicks | 200 ms or less | Over 500 ms | Heavy JavaScript, page builders, chat widgets and third-party scripts |
| CLS (Cumulative Layout Shift) | How much the layout jumps while loading | 0.1 or less | Over 0.25 | Images without set sizes, late-loading banners, font swaps |
Reading PageSpeed Insights line by line
Run the test on the mobile tab first; phones are how many Thai customers browse. Then read the report from top to bottom in this order.
- The field data panel. At the top, "Discover what your real users are experiencing" shows CrUX data for the exact URL and for the whole site (origin). A "Passed" or "Failed" Core Web Vitals assessment here is the headline, not the score. Small sites may see no field data at all.
- The performance score. A 0 to 100 weighted summary of lab metrics. Use it to compare before and after a change on the same page, not to compare your site with a competitor's.
- Lab metrics. LCP, CLS, Total Blocking Time (a lab stand-in for responsiveness), First Contentful Paint and Speed Index.
- Diagnostics and insights. The list of specific problems, such as oversized images, unused JavaScript or render-blocking resources, often with the exact files named. This is the useful part.
Expand each diagnostic and note which files it names. A plugin or theme name appearing again and again tells you where the weight comes from.
GTmetrix and WebPageTest
Both tools run lab tests like Lighthouse, but add controls PageSpeed Insights lacks. The most important is test location. If your customers are in Thailand, choose a server in Asia where the tool offers one; a test from North America adds distance your visitors never experience and makes a Bangkok-hosted site look slower than it is. Pick a mobile device profile and connection speed too.
Their best feature is the waterfall: a timeline of every file the page downloads, in order. Read it for three things: a long first bar (the server is slow to respond), many files from other domains (third-party scripts), and very long bars for single images or fonts. If the static files come from far away, Cloudflare CDN and security can serve them from closer locations.
The Core Web Vitals report in Search Console
If you have not connected your site yet, follow our Google Search Console setup guide first. The report sits under Experience, Core Web Vitals, with separate mobile and desktop views.
Statuses
Each URL is marked Good, Needs improvement or Poor, based on field data. A URL's status is set by its weakest metric: good LCP and CLS with poor INP means Poor.
URL groups
Search Console does not list every page individually. It groups similar pages, such as all product pages or all blog posts, and reports the group's figures with example URLs. That is useful: one template fix usually fixes the whole group.
Not enough data
Small business sites often see empty charts or a message that there is not enough usage data. CrUX only reports when enough real Chrome visits exist, so a new or low-traffic site may have none. That is not a penalty. Use lab tests in the meantime and aim for good lab numbers on key pages.
From diagnosis to fix
Most diagnostics fall into a few families. Here is where each usually leads; the full list of causes and repairs for WordPress is in our guide on why a WordPress site is slow.
- Slow server response or long first waterfall barHosting, PHP version or missing page caching.
- Largest Contentful Paint image is largeResize, compress and serve WebP; do not lazy-load the hero image.
- Reduce unused JavaScript or CSSPlugins or a page builder loading code on every page.
- Poor INP or high Total Blocking TimeHeavy scripts: chat widgets, pixels, sliders, builder animations.
- Layout shiftsSet width and height on images and embeds, reserve space for banners, preload key fonts.
- Large font filesFewer weights, and subset Thai and English fonts.
How do you validate a fix?
After you fix an issue, re-run the lab test on the same page and device settings, and compare with your saved screenshot. If the lab numbers improved, open the issue in the Search Console Core Web Vitals report and choose Validate fix. Search Console then monitors the affected URLs using field data over a 28-day period, so expect the result to take around a month, not a day. If some URLs still fail, validation stops and you fix and try again.
Keep a note of what changed and when, so you can link the eventual result to the right fix. For more guides on performance and measurement, browse our SEO, speed and analytics articles. If you would like help, our speed work starts with exactly this kind of reading, and if the answer is a new site, request a website development quote and we will scope it with you.
More speed and search guides
Sources
- web.dev: Web Vitals
- Search Console Help: Core Web Vitals report
- PageSpeed Insights: About
- web.dev: INP becomes a Core Web Vital on March 12
Builds and ranks the websites: the systems behind the 17 client sites whose Search Console numbers we publish. This guide was written by the Website Development Bangkok team. More about us