PageSpeed Insights

Paste a URL at pagespeed.web.dev or call the PSI API. The field section shows LCP, INP, and CLS at the 75th percentile over the trailing 28 days, if the URL or origin has enough Chrome traffic. The lab section runs Lighthouse on an emulated mid-tier phone with throttled network and returns a score from 0 to 100.

The score is a weighted blend of lab metrics, not of field data. The pass or fail Core Web Vitals assessment at the top uses field data only. If CrUX has no data for the URL, you see no field block and only the lab result.

A bakery's Shopify store shows a mobile score of 41 and a field assessment of Passed, with LCP 2.2 s, INP 160 ms, and CLS 0.04. The owner sees a red 41 and worries. Lab LCP is 6.1 s because the test emulates slow 4G with 4x CPU slowdown, while real customers mostly sit on faster phones.

The team removes a review widget that adds 380 KB of script. The lab score rises from 41 to 58. The field numbers move only as the 28-day CrUX window rolls, so the top block does not change the next morning.

PageSpeed Insights is a web report. Lighthouse is the engine behind its lab half, and also runs in DevTools, the command line, and CI. PSI adds CrUX field data and runs from Google's servers; Lighthouse in your own DevTools runs from your machine and your network.

Chasing the 0 to 100 score instead of the vitals. Running it once and trusting the result when scores swing by several points. Reading origin-level CrUX data as if it belonged to the single URL. Expecting field numbers to react within a day. Testing a page behind a login, which PSI cannot reach.

Use it as a quick check on public URLs: read field vitals first, then the Opportunities and Diagnostics lists for fixes. Run it three times and take the median lab score. For ongoing tracking, call the API from a script and store results weekly instead of refreshing the page by hand.