
Core Web Vitals: LCP, INP and CLS and their thresholds
8 min read below · WebYes knowledge base
Last updated:
What do LCP, INP and CLS measure, which values are good and what changed with INP in 2024? Explained with thresholds and Google sources.
Core Web Vitals are three Google measurements of user experience: LCP (loading, good ≤ 2.5 s), INP (response to interaction, good ≤ 200 ms) and CLS (stability, good ≤ 0.1), assessed at the 75th percentile. INP replaced First Input Delay as a Core Web Vital on 12 March 2024.
The three metrics and their thresholds
Largest Contentful Paint (LCP) measures how long it takes for the largest visible element (usually an image or heading) to appear. Interaction to Next Paint (INP) measures how quickly the page responds to clicks, taps and key presses. Cumulative Layout Shift (CLS) measures how much the layout shifts while loading, for instance when a banner pushes the text down.
| Metric | Good | Needs improvement | Poor |
|---|---|---|---|
| LCP (loading) | ≤ 2.5 s | 2.5 - 4 s | > 4 s |
| INP (interaction) | ≤ 200 ms | 200 - 500 ms | > 500 ms |
| CLS (stability) | ≤ 0.1 | 0.1 - 0.25 | > 0.25 |
Google assesses the 75th percentile of page visits: slower devices and connections need to score well too.
The thresholds come from Google's official documentation on web.dev. INP replaced First Input Delay as a Core Web Vital on 12 March 2024, as Google announced on web.dev. The values apply per page, not per site: your homepage can score well while a product page fails. A site that feels smooth in the office can still miss the threshold on an average phone over 4G.
Why Core Web Vitals matter
Slow or stuttering pages feel like friction to visitors, and that can mean people leaving and fewer conversions. Google uses the Core Web Vitals in its ranking systems; the relevance of the content still weighs heavier. For web shops it is extra relevant, because a stuttering checkout can directly affect revenue.
The metrics are deliberately chosen around what a visitor actually notices: how quickly do I see something, how quickly does it respond, and does the layout stay put. A server that responds fast but a page that then spends seconds running scripts rightly scores poorly. This is not about an abstract grade; it is about how the page feels in the user's hand.
Lab data versus field data
There are two kinds of measurements. Field data comes from real visitors, for example through the Chrome User Experience Report, and shows what they actually experience. Lab data comes from a test in a controlled environment, such as a scan or Lighthouse run, and is instantly available and reproducible. Web.dev advises giving field data priority; lab data remains the tool for finding problems and verifying fixes.
The WebYes scan checks technical signals that influence the Core Web Vitals on the scanned pages (up to six per scan) and includes them in the speed pillar. LCP, INP and CLS themselves are not a scan result. To learn how to measure and interpret them yourself, also read testing website speed. For a quick external check you can also use PageSpeed Insights; it often shows both lab and field data when there is enough traffic.
The most common causes of poor scores
A high LCP often comes from a slow server response, a largest element that is found or loaded too late (for example through lazy loading of the largest image), oversized files or render-blocking scripts and stylesheets. A high INP often comes from JavaScript keeping the main thread busy with long tasks. CLS problems come from images or videos without fixed dimensions, ads and widgets that resize themselves, or web fonts that make the text jump.
A practical order of attack is: optimise images first (size, compression, lazy loading), then remove or defer unnecessary scripts, and finally speed up the server side with caching or a CDN. Measure again after every step; without a new measurement you do not know whether the fix really works.
How WebYes includes Core Web Vitals in the keurmerk
Speed is one of the four pillars of the WebYes scan, alongside mobile, security and accessibility. Within that speed pillar we look at, among other things, server response time, render-blocking CSS and scripts, the largest image (priority and lazy loading), images without dimensions and the number of external scripts: signals that influence LCP, INP and CLS. The free scan is labelled beta and limited to a maximum of six pages per run; a paid full audit may include up to fifteen sitemap-led pages.
The keurmerk requires an average score of at least 80 across the four pillars, and each pillar separately must score at least 60. A site with a strong mobile score but a speed pillar below 60 therefore fails the assessment, even if the average is high. That is intentional: a keurmerk that leans on only one strong pillar says little about the experience of your visitors. More on how to measure and prioritise yourself is in testing website speed and in the context of mobile-friendly websites, where slow LCP and CLS often hit hardest.
Practical order: what to fix first on real sites
A practical order of wins is this. Start with LCP on the homepage and key landing pages: hero images in a modern format (WebP or AVIF), fixed width/height, and no render-blocking CSS delaying the largest paint. Measure again and check whether the LCP gets under 2.5 seconds.
Then INP: inventory third-party scripts (chat, A/B tools, pixel stacks) and disable what is not essential on first paint. Chat widgets and other scripts that keep the main thread busy can worsen the INP. Only then fine-tune: code-splitting, idle callbacks, and less work in event handlers. CLS often comes last: reserve space for banners and cookie bars so content does not jump when consent UI appears.
Document every fix with a before/after lab measurement on the same URL. Without that discipline you optimise blindly and cannot tell which change saved the pillar score. The score chart on a domain's WebYes profile shows how the total score develops; so record the speed pillar score per scan yourself.
What Core Web Vitals do not measure
Core Web Vitals say nothing about content quality, accessibility or security. A fast page can still fail on contrast, missing alt text or missing security headers. That is why WebYes scores four pillars separately: speed alone is never enough for the keurmerk.
They also do not measure conversion paths end to end. A checkout that scores well in the lab but has three confusing steps can still lose visitors. Use vitals as a hard floor for technical UX, not as a substitute for usability tests or a human WCAG audit.
Sources
- Web Vitals (LCP, INP, CLS) - web.dev (consulted on 13 October 2026)
- INP wordt Core Web Vital op 12 maart - web.dev (consulted on 13 October 2026)
- Page experience in Google Search - Google Search Central (consulted on 13 October 2026)
- Largest Contentful Paint (LCP) - web.dev (consulted on 13 October 2026)
- Interaction to Next Paint (INP) - web.dev (consulted on 13 October 2026)
- Cumulative Layout Shift (CLS) - web.dev (consulted on 13 October 2026)
Frequently asked questions
Are Core Web Vitals an official ranking factor?
Google uses the Core Web Vitals in its ranking systems, but it is not a single signal and a good score does not guarantee a top position. Google always shows the most relevant content, even when page experience is mediocre; for pages that are both relevant, a good page experience can contribute to success in Search. Content and relevance weigh heavier; speed is not a substitute for a weak page.
What happened to First Input Delay (FID)?
Since 12 March 2024, First Input Delay (FID) has been replaced as a Core Web Vital by Interaction to Next Paint (INP). INP looks at the response time of all interactions on a page, not just the first, and so gives a more realistic picture.
Does the WebYes scan measure Core Web Vitals?
Not directly. Speed is one of the four pillars. The scan checks technical signals that influence the Core Web Vitals, such as server response time, render-blocking files and images without dimensions, on up to six pages in the free scan (beta) and up to fifteen in the full audit, and shows per finding where the delay comes from. Google measures LCP, INP and CLS themselves on real visitors (field data); you can see them in PageSpeed Insights, for example. For the keurmerk: average ≥ 80 and each pillar ≥ 60.
Why do lab and field scores sometimes differ?
Lab tests run on a fixed, simulated connection and a fixed device profile. Field data is the 75th percentile of real visitors, with all the differences in devices and networks. A lab score that is just good can still look mediocre in the field. Use lab to debug; give field data priority when setting priorities.
Which pages should I optimise first?
Start with pages that have traffic and conversion: homepage, product or service pages, and checkout. Optimising a rarely visited archive page barely lifts your field percentile. In the WebYes full audit the crawler picks up to fifteen sitemap-led URLs; that sample is a good starting list for what to tackle first.
The WebYes scan measures this too
Scan your website for free on speed, security, mobile and accessibility and see where you stand.
Start free scan

