What does a website speed test measure?
The time a server takes to respond is only part of how fast a page feels. Visitors also wait for content to appear, interact with controls and experience any unexpected movement while the page loads. A useful performance review keeps those observations separate.
The source audit records request duration including DNS and download time, plus the HTML response size. These are diagnostic observations rather than a browser loading-speed score. Where configured, the full report adds Google Lighthouse mobile and desktop tests for the submitted page.
Lighthouse lab tests and real visitors answer different questions
Lighthouse runs a simulated test. The report can include its performance score and measurements such as Largest Contentful Paint, First Contentful Paint, Cumulative Layout Shift, Total Blocking Time and Speed Index. Mobile and desktop results are kept separate because their test conditions differ.
Google CrUX is based on eligible real Chrome visitors and may include LCP, Interaction to Next Paint and CLS at the 75th percentile. Smaller websites may not have enough public data. The report labels missing data and any whole-origin fallback rather than inventing a measurement for the exact page.
Where to begin if your page feels slow
Check which device, URL and date the measurement covers before comparing scores. Review the saved Lighthouse findings for resources that delay rendering or images that could be delivered more efficiently. Test an important landing page as well as your homepage.
For example, a large hero image may be a useful place to investigate if the main content appears late. Ask your developer to check the rendered image size, compression, loading priority and caching. Confirm the actual bottleneck rather than installing several optimisation plugins in response to one score.
Why performance results can change
Lab scores can vary with test conditions, page content and third-party resources. Real-user measurements describe their stated collection period and a different population. They should not be treated as a direct before-and-after comparison without checking those differences.
A one-off report preserves its saved results and timestamps. It is not continuous speed monitoring. Successful checks are cached for that report; use a new audit when you need a new snapshot after website changes.
Turn the evidence into a useful next step
| Signal | What to review |
|---|---|
| LCP · loading | How quickly the main content appears. Investigate the actual resource and rendering work behind a slow result. |
| INP · interaction | A real-user responsiveness measurement when CrUX has sufficient data. It is not the same as lab Total Blocking Time. |
| CLS · stability | Unexpected movement during page loading. Review image dimensions, late content and font behaviour. |
Google: understanding Core Web Vitals ↗
Use the report’s evidence and scope alongside the relevant documentation for your website. Automated findings are a starting point for investigation.
Questions about this check
Is the free overall score a PageSpeed score?
No. The free overall score uses seven categories of source checks. Lighthouse speed scores and CrUX metrics appear separately in purchased reports when configured and completed.
Why is real-user performance unavailable for my website?
Google needs sufficient eligible traffic to publish CrUX data. A lack of public data is not a speed failure or a passing result.
Does this test every page or location?
No. Enhanced performance checks cover the submitted page, with origin-level real-user data explicitly labelled if used. This is not a multi-location, full-site performance test.
All seven audit categories · How the overall score is calculated · What to fix first