Core Web Vitals have become essential ranking signals — but more importantly, they measure real user experience. Here's what matters in 2024 and how to actually improve your scores.
The Three Metrics That Matter
Largest Contentful Paint (LCP)
LCP measures how quickly the largest visible element loads. Target: under 2.5 seconds. The most common culprits are unoptimised hero images, slow server response times, and render-blocking resources.
- Serve images in modern formats (WebP, AVIF)
- Use a CDN for static assets
- Add
fetchpriority="high"to your LCP image - Reduce Time to First Byte (TTFB) at the server level
Interaction to Next Paint (INP)
INP replaced FID in March 2024. It measures responsiveness across all interactions, not just the first. Target: under 200ms. Heavy JavaScript is usually the cause.
- Break up long tasks with
scheduler.yield() - Defer non-critical JavaScript
- Audit and remove unused third-party scripts
Cumulative Layout Shift (CLS)
CLS measures visual stability — how much the page unexpectedly moves while loading. Target: under 0.1. Ads, images without dimensions, and late-loading fonts are common culprits.
- Always set explicit width and height on images and videos
- Reserve space for ad slots with CSS
- Use
font-display: optionalto prevent font-related shifts
How to Measure
Field data (real user measurements) is what Google uses for ranking. Lab data (simulated tests) is what you use to diagnose problems. Use both:
- Google Search Console — field data for your whole site
- PageSpeed Insights — both field and lab data per URL
- Chrome UX Report — raw field data via BigQuery
- Lighthouse — detailed lab diagnostics in DevTools
Prioritising What to Fix
Start with pages in the "Poor" threshold (not "Needs Improvement") — those are the ones actively costing you. Use Search Console's Core Web Vitals report to identify which URL groups are underperforming, then use PageSpeed Insights to diagnose the specific cause.
A common mistake is chasing lab score improvements without checking whether field data actually improves. Always validate with real-user data.