You’re probably using Lighthouse wrong bt Filip Rakowski
summary
Argues that Lighthouse is useful but often misused because its lab score is synthetic, algorithm weights change, and local runs vary by CPU and extensions; recommends using Lighthouse CI in CI to detect regressions, reading field data from PageSpeed Insights which surfaces Core Web Vitals from the Chrome User Experience Report, and checking WebPageTest for actionable traces; cites Moto G4 emulation and the Adidas case where an 80+ Lighthouse score still maps to terrible real‑user performance, so prioritize real user metrics over a single score.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:07Web performance in e-commerce
- 6:22Lighthouse rise and limits
- 8:47What Lighthouse measures
- 13:21How Lighthouse score is calculated
- 15:49Local Lighthouse runs unreliable
- 17:37Best Lighthouse runs
- 20:06PageSpeed Insights real data
- 22:12Tricking Lighthouse misuse
- 24:41Lighthouse flaws and usefulness
- 26:29Lighthouse in CI pipelines
- 27:55Real UX vs synthetic scores
- 31:02Summary
related talks
22mYou’re Probably Using Lighthouse Wrong
38mYou’re probably using Lighthouse wrong
42mYou’re probably using Lighthouse wrong
27mCommon misconceptions around Google Lighthouse
26mDemystifying Lighthouse, CrUX, and Core Web Vitals
27mHow to Measure Performance Effectively?
16mFrom beginner to expert in web performance
35mExploring the Lighthouse