The Core Web Vitals are the set of metrics Google uses to measure a user's real experience on a site. In 2026, three indicators define the game: LCP (Largest Contentful Paint), INP (Interaction to Next Paint), and CLS (Cumulative Layout Shift). Sites that fail on these metrics lose ranking, even with excellent content and strong backlinks.
This guide shows what each metric measures, what the acceptable thresholds are, how to diagnose problems, and which engineering actions deliver real gains, not a placebo.
What each Core Web Vital really measures
LCP (Largest Contentful Paint) measures the time until the largest visible element in the viewport is rendered. It is "when the user perceives that the page has loaded". Good: up to 2.5s. Needs improvement: 2.5s to 4s. Poor: above 4s.
INP (Interaction to Next Paint) replaced FID in 2024 and measures the delay between a user interaction (click, tap, keyboard) and the next visual render. It is the hardest metric to optimize because it depends on the main thread being free. Good: up to 200ms. Needs improvement: 200ms to 500ms. Poor: above 500ms.
CLS (Cumulative Layout Shift) measures the sum of unexpected element shifts during loading. That banner that shifts the button and makes you click wrong is high CLS. Good: up to 0.1. Needs improvement: 0.1 to 0.25. Poor: above 0.25.
How Google actually uses these metrics
Google evaluates Core Web Vitals based on CrUX (Chrome User Experience Report), that is, field data collected from real Chrome users, not lab data from Lighthouse. This means the metric that matters is the 75th percentile of real users over the last 28 days, segmented by device.
To be eligible for the Page Experience boost, all three metrics need to be in "good" at the 75th percentile. Just one poor metric already disqualifies the URL. That is why it is common for sites with a Lighthouse score of 95+ to still have ranking problems: Lighthouse runs in a controlled condition, while CrUX runs in the real world, with 3G, weak hardware, and heavy tabs.
Optimizing LCP: the 5 highest-impact actions
1. Identify the LCP element with the Chrome Performance tab. It is usually a hero image, video, or large text block. 2. Serve the critical asset early: preload in the link tag, fetchpriority="high" on the LCP image, and remove render-blocking CSS/JS. 3. Use modern formats: WebP or AVIF for images, with a responsive srcset. 4. Aggressive CDN: edge caching with a high TTL for static assets. 5. Fast server response: TTFB below 600ms; HTML caching when possible, with optimized SSR.
INP: the metric that most brings sites down in 2026
INP is where most superficial optimizations fail. The culprit is almost always heavy JavaScript blocking the main thread. Effective strategies: aggressive code splitting, loading only what is needed per route; defer and async on non-critical scripts; web workers for heavy processing off the main thread; debounce and throttle on frequent handlers (scroll, input); removal of heavy libraries used for trivial tasks (replacing Moment.js with date-fns or native Intl, for example).
Modern frameworks like Next.js, Remix, SvelteKit, and Astro are born with practices for minimizing client-side JS. Migrations from old stacks to these frameworks usually deliver an INP gain of 40 to 70% in a few weeks.
CLS: common causes and definitive solutions
High CLS almost always comes from five sources: images without declared width/height (solved by reserving space in the CSS or the HTML attribute); web fonts without proper font-display (use swap or optional); cookie banners, ads, and offers that appear after loading and push content (reserve fixed space); dynamic content injected by A/B testing without a placeholder; animations that change layout instead of using transform and opacity (which do not cause reflow).
Tools you should use
The essential kit is PageSpeed Insights (combines CrUX and Lighthouse), Search Console > Core Web Vitals (a per-URL view, grouped by pattern), WebPageTest (reproducible tests with a detailed waterfall), Chrome DevTools Performance (profiling in development), and RUM tools like Vercel Analytics, Speedcurve, or Sentry Performance for continuous monitoring in production.
Real business impact
Optimizing Core Web Vitals has a direct impact on three fronts: SEO (a ranking boost when all metrics are green), conversion (every 100ms less in LCP usually raises conversion by 1 to 3% in e-commerce), and retention (slow sites have a bounce rate up to 90% higher on mobile). Performance became a product metric, not an IT one.