What PageSpeed Insights measures
The report has two halves and they come from different places. The top half, labelled as real user data, is drawn from the Chrome User Experience Report: measurements collected from actual Chrome visitors over a recent rolling window. It shows the three Core Web Vitals — Largest Contentful Paint, Interaction to Next Paint and Cumulative Layout Shift — alongside first contentful paint and time to first byte, split between mobile and desktop.
The bottom half is a Lighthouse audit run on demand: one simulated load on a throttled connection and a modest device, producing the performance score and the list of opportunities and diagnostics. Because it is a single simulation rather than a record of real visits, it is a diagnosis. It tells you what is heavy or badly ordered on the page, not what your audience experienced.
Why PageSpeed Insights matters
It is the easiest place to see the gap between how a site feels on your own machine and how it behaves for the people who actually use it. A site built and reviewed on a fast office connection can be slow and jumpy on a mid-range phone. In Nepal, where most traffic arrives on mobile and connection quality varies a great deal between places and times of day, that gap is usually wider than anyone expects.
It also hands a developer a prioritised list instead of a complaint. “The site is slow” starts an argument. A named render-blocking script with an estimated saving next to it starts a task.
Common mistakes with PageSpeed Insights
The most expensive is optimising the score rather than the experience. The score comes from the lab run, while Google’s page experience signals use the field data. Work that lifts the number without moving Core Web Vitals for real visitors has bought you a screenshot, not a faster site.
The second is testing only the homepage. Homepages tend to be the lightest and best-maintained page a site has; the template that matters is the product page, the article or the landing page people actually arrive on. The third is reacting to one run. Scores wobble between tests because conditions vary, so a change is only real when it survives repeated testing.
How to act on it
Start with the field data if the page has any, because that is what Google assesses and what your visitors lived through. Work out which of the three metrics is failing and treat it as its own problem: Largest Contentful Paint is usually a hosting, image or font issue, Cumulative Layout Shift is a layout and ad-slot issue, and Interaction to Next Paint is usually JavaScript.
Then use the lab half to find the cause, change one thing, and re-test the same URL. Resist bundling several fixes together, because you learn nothing about which one worked. Where field data stays poor after the obvious wins, the constraint is often the server or the theme itself, and that becomes a page speed optimisation project rather than a plugin setting.
