Guides
How to put Lighthouse audits somewhere useful — a CI step, a scheduled check, a performance budget that fails the build.
-
Do Lighthouse scores change when you test from a different region?
We ran one page from seven regions. The measured round trip went from 2.5 ms to 275 ms and the score stayed at 100. Here is what region really changes. -
How to monitor Lighthouse scores over time
Scheduling the runs takes an afternoon. Knowing which movement is a real regression is the hard part — here is what 41 runs of one page say about it. -
How to run Lighthouse in CI
Lighthouse CI takes ten minutes to set up. Then two defaults decide what your build really checks — one run, and the best one. -
How to run Lighthouse in GitLab CI
A working .gitlab-ci.yml, plus the trap in the standard image — lighthouse gives you 13.4.1 and lhci gives you 12.6.1 in the same job. -
How to run Lighthouse in a GitHub Action
The workflow takes five minutes. Then the runner you get decides what the numbers mean — we measured five runs of one page at 32, 34, 34, 59 and 57. -
PageSpeed Insights API rate limits and quota exceeded errors
Google publishes no quota figure for the PageSpeed Insights API. Here is what the 429 payload tells you, where to read your real limit, and when to retry. -
What is a good Lighthouse score?
Google calls 90 and above good. The median mobile page scores 49. What "good" means for each of the four categories, and what to aim at instead. -
Why are my Lighthouse scores different?
Two problems hide in that question — run-to-run noise and a configuration offset. How to tell them apart, and what 130 fixed-rig runs show. -
Why your site scores differently in another country
Distance and device speed explain most regional score gaps. Sometimes the page itself is different. Two measured cases, from 2,800 Lighthouse runs.