Google PageSpeed Insights

Four Lighthouse scores per strategy — performance, accessibility, best practices and SEO — on the usual 0 to 100 scale. Beneath them, the lab metrics that produced the performance score: First Contentful Paint, Largest Contentful Paint, Total Blocking Time, Cumulative Layout Shift and Speed Index.

Then Lighthouse’s own warnings about the run, the opportunities it suggests — ordered by what it estimates each is worth — a bounded set of diagnostics describing the page, and a collapsed list of what passed.

Then, separately, field data from the Chrome UX Report: what real Chrome users actually experienced over the last 28 days. Google publishes page-level data only where there is enough traffic to report without identifying anybody; where there is not, this tool falls back to the whole origin’s data and says so in as many words. A fast page can sit behind a slow site average, and the two are never shown as the same thing.

A lab score is one page load, on Google’s hardware, under simulated network and CPU conditions. It varies between runs — sometimes by ten points — and it is a diagnostic: it tells you what to fix, in a controlled setting where the answer is reproducible enough to act on.

Field data is the opposite trade. It is what happened to real people on real devices and real connections, which is what actually matters and what Google uses for Core Web Vitals — but it lags by weeks and cannot tell you which change caused what. This tool keeps the two apart and labels both, because averaging them in your head is how people end up optimising for the wrong one.

• ·A category Lighthouse could not score is shown as not measured, never as a zero. An absence of data and a bad result are different things.
• ·Nothing is re-scored, re-weighted or combined into a SearchOrbit AI number. The scores are Google's, on Google's scale, and the Lighthouse version that produced them is on the result.
• ·Results are cached for 30 minutes per page. A cached answer is labelled as one and shows when it was measured — it is never presented as newly taken.
• ·Field data comes from the slice PageSpeed includes, page-level where Google has it and whole-site where it does not — always labelled as one or the other. The dedicated Chrome UX Report API returns more, and moving to it is later work rather than something this tool quietly does today.
• ·Opportunities and diagnostics are the highest-impact handful, not everything Lighthouse found. Google's own PageSpeed Insights page lists them all, and the result says when there were more.
• ·Only public http:// and https:// addresses on the standard ports are measured. Nothing is sent to Google until the address has passed the same validation every other tool here applies.
• ·This is the smallest allowance in the library, because each check costs two calls against a Google quota shared by everyone using SearchOrbit AI rather than a per-visitor one.

Speed is one signal. For what a crawler reads on the same page, run a free SEO audit, or check the meta tags it publishes.