Your Lighthouse Score Doesn’t Matter on 3G
summary
Lighthouse and DevTools lab runs are not the whole story because throttling only simulates network not low CPU on real entry tier phones so a 100 Lighthouse score still leaves users waiting on 3G and $100 Android hardware where roughly 39% of users in some regions live and only 27% use flagships. Practical steps: buy a $100 Android phone, track RUM and device cohorts, run bundle analysis, audit runtime on low‑end Android, prioritize server rendering, selective hydration, critical CSS and a send‑less execute‑less budget as a delivery requirement.
AI-written summary · about
chapters (13)
- 0:00Why Lighthouse scores don’t tell whole story
- 1:58Performance metrics fail real users
- 4:27Test on premium, users on cheap Androids
- 6:443G throttling in DevTools vs real devices
- 8:50Browser pressure on low-end hardware
- 10:39Hydration delays hide non-interactive UI
- 12:03Bloat from unnecessary dependencies
- 13:48Auditing beyond Lighthouse metrics
- 15:33Testing with others reveals issues
- 16:39Critical CSS and server rendering
- 18:04Defer non-critical actions
- 19:08Performance in SDLC
- 20:14Summary and call to action
related talks
27mCommon misconceptions around Google Lighthouse
22mYou’re Probably Using Lighthouse Wrong
16mFrom beginner to expert in web performance
40mYou’re probably using Lighthouse wrong bt Filip Rakowski
42mYou’re probably using Lighthouse wrong
26mDemystifying Lighthouse, CrUX, and Core Web Vitals
38mYou’re probably using Lighthouse wrong
19mLighthouse for mobile apps