Core Web Vitals are among the page experience signals Google takes into account in its ranking systems. They do not decide rankings on their own: content relevance comes first. But when relevance is comparable, a fast, stable page has an edge, and it gives visitors a better experience. Since March 2024, INP (Interaction to Next Paint) has replaced FID; the thresholds Google publishes have not changed. This technical guide explains how to measure, diagnose and optimize each metric.
The 3 Core Web Vitals metrics in 2026
LCP (Largest Contentful Paint) — Threshold: ≤ 2.5 seconds
LCP measures how long it takes to render the largest visible element in the viewport. It is the metric that best reflects how users perceive loading speed. Google rates an LCP of 2.5 seconds or less as "good," anything above as "needs improvement," and above 4 seconds the experience is rated "poor."
The most common causes of a slow LCP:
- Unoptimized images: JPEG/PNG instead of WebP/AVIF, files too heavy for their display size, or a main image loaded lazily, which delays its display
- Render-blocking resources: CSS and JavaScript that delay the initial render
- Server response time (TTFB): A slow server or a poorly configured CDN
- Web fonts: Custom fonts that block text from displaying (FOIT)
INP (Interaction to Next Paint) — Threshold: ≤ 200 ms
INP replaced FID in 2024 and measures the overall responsiveness of the page throughout the visit — not just the first interaction. It is more demanding than FID: it observes clicks, taps and key presses, and keeps one of the slowest interactions of the visit.
The most common causes of a high INP:
- Heavy JavaScript on the main thread: Third-party scripts (analytics, chat, ads) that hog the CPU
- Framework hydration: React, Next.js and similar frameworks run a lot of JavaScript on load to make the page interactive; an interaction during this phase has to wait
- Expensive event handlers: Complex calculations triggered by clicks or scrolls
- Costly rendering after the interaction: a very large DOM or forced layout recalculations delay the next paint
CLS (Cumulative Layout Shift) — Threshold: ≤ 0.1
CLS measures the visual stability of the page. Every time an element moves unexpectedly (an image that loads and pushes the text down, an ad banner that gets inserted), the CLS score goes up. A CLS above 0.1 "needs improvement," and above 0.25 it is "poor."
Measuring: field data and lab data
Google assesses Core Web Vitals on field data: the Chrome User Experience Report (CrUX), collected from real Chrome users over the last 28 days. A page passes the assessment when all three metrics are "good" at the 75th percentile of visits. The Search Console Core Web Vitals report shows this data by groups of URLs; PageSpeed Insights shows it for a single URL, next to a Lighthouse lab test.
Lab tests (Lighthouse, Chrome DevTools) are for diagnosing a problem and checking a fix before it goes live. They do not replace field data. A standard Lighthouse test does not measure INP, which requires real interactions: Total Blocking Time (TBT) gives an indication instead.
Optimizing: the levers for each metric
- LCP: serve the main image in a modern format and at the right size, preload it or give it a high loading priority (fetchpriority attribute), and reduce TTFB (caching, CDN) and render-blocking CSS
- INP: break up long tasks and regularly yield to the browser, defer or slim down third-party scripts, and limit the work triggered by each click and the size of the DOM
- CLS: give images, videos and iframes dimensions (width and height, or aspect-ratio), reserve space for ads and banners, and choose fallback fonts close to the final font
How Zerkado measures your Core Web Vitals
Zerkado's technical SEO audit measures LCP, CLS and TBT in the lab on a sample of pages, in a real browser with a mobile viewport, as well as the response time of every crawled page. For each issue found, it explains how to fix it. INP requires real visitors: TBT gives a lab approximation, and field data remains available in Search Console.
The SEO audit module ranks issues by severity (critical, warning, information) and by the number of pages affected, to help you decide what to fix first. Each new crawl of the site updates the measurements: a page that slows down after a release shows up on the next pass.
To go further with your site's technical optimization, read our complete guide to technical SEO audits and our article on the most common technical SEO mistakes.