Core Web Vitals: what still matters
Three years in, only some of the metrics still matter for actual outcomes. A short prioritization.
Core Web Vitals are now table stakes for most projects. Three years of data later, only some of the metrics actually move outcomes.
LCP: still matters
Largest Contentful Paint correlates with bounce rate in every dataset we've worked on. A slow LCP is a real problem. Fix this first.
CLS: still matters
Cumulative Layout Shift annoys users in obvious ways. Modern frameworks make CLS easy to avoid - explicit image dimensions, reserved space for ads - but only if the team remembers.
INP: matters more than FID did
Interaction to Next Paint replaced First Input Delay because FID was easy to game. INP measures real-world interaction lag. If your buttons feel sluggish, INP will catch it.
TBT: matters as a debugging aid
Total Blocking Time isn't a Core Web Vital but it's the metric we look at first when LCP or INP regresses. It points at the JS work that's blocking the main thread.
What doesn't matter as much as you think
First Contentful Paint. Speed Index. The other Lighthouse numbers. They're useful as diagnostic signals but don't move SEO or conversion the way LCP and INP do.
Where to focus
Optimize images, reduce JS, eliminate layout shifts on initial load. That trifecta covers 90% of vitals fixes. Everything else is a long tail that may or may not be worth the engineering time.