On this page
Core Web Vitals are not measured in a lab. Google collects them from real Chrome users over the past 28 days and looks at the 75th percentile, not the average: a page passes only when three in four visitors get a good experience, including the one on an old phone on a weak 4G signal. That is why a site that feels quick on the office Wi-Fi can still show red in Search Console.
Google introduced the three metrics in May 2020 and has used them in search since June 2021. In March 2024 INP replaced FID, which only measured the first click. Google is clear that good scores do not guarantee rankings and relevant content comes first. But between two equally useful pages the faster one wins, and a slow page mostly loses visitors before they read a word.
The three numbers and their limits
| Metric | What it measures | Good | Poor |
|---|---|---|---|
| LCP, Largest Contentful Paint | When the largest image or text block appears | Up to 2.5 s | Over 4 s |
| INP, Interaction to Next Paint | How fast the page responds to a click or tap | Up to 200 ms | Over 500 ms |
| CLS, Cumulative Layout Shift | How much content jumps while loading | Up to 0.1 | Over 0.25 |
Everything in between counts as "needs improvement". Your own figures are in Search Console under Core Web Vitals, and per page in our website speed test.
A heavy home page, before and after the work
LCP Main content loadedheavy homepage 5.2s / optimised 1.8s
goodneeds workpoor
INP Response to a clickheavy homepage 420ms / optimised 120ms
goodneeds workpoor
CLS Layout shiftheavy homepage 0.31 / optimised 0.02
goodneeds workpoor
Google's thresholds at the 75th percentile. The two values per gauge are an example, not a measurement.
What a tenth of a second is worth
In 2020 Deloitte and Google measured what faster mobile sites did for 37 brands. One tenth of a second faster meant on average 8.4 percent more conversions in retail and 10.1 percent in travel.
Speed pays twice in paid search: landing page experience feeds your Quality Score and therefore your cost per click, and a faster page converts more of the visitors you paid for; see conversion rate optimisation.
Where the seconds usually hide
- The hero imageA 4 MB photo at the top is the classic cause of a slow LCP. As WebP or AVIF at the right width it weighs a tenth.
- Third-party JavaScriptPage builders, sliders, chat widgets and tracking scripts make INP sluggish. Every script your visitor does not need should go.
- Images without dimensionsWithout width and height, text jumps when the image loads. That is CLS, and it causes misclicks.
- A distant serverShared hosting on another continent costs every visitor half a second before anything happens. A static site on a CDN near your buyers does not.
For sites selling into several countries the last point matters most: your buyers in Madrid, Munich and Manchester should all be served from nearby. That is one reason we build multilingual websites as static sites on a CDN.
A villa site, seven megabytes lighter
When we rebuilt the site of Villa Lidia, a holiday villa in Umbria, the home page weighed 8.4 MB, mostly in photographs. After the rebuild it weighs 1.4 MB, and the image library went from 51 MB to 17 MB without a single photo looking worse. Every image now exists in four widths as WebP, the browser picks the smallest that fits, and every image declares its size so nothing jumps.
On a page builder or a heavy WordPress theme, part of the slowness sits in the foundations; see why WordPress is so slow. The step-by-step fixes for an existing site are in our guide to improving website speed.
Do not chase a perfect lab score
Do not chase 100 in PageSpeed Insights. That is a lab test on a simulated phone. Google ranks on the field data in Search Console, which only appears once a page has enough visitors.
Speed questions from site owners
Why do PageSpeed and Search Console disagree?
PageSpeed shows a lab test on a simulated phone and, with enough traffic, field data too. Search Console shows only real visitors over 28 days.
Are Core Web Vitals a ranking factor?
Yes, but a small one. Relevant content weighs more. Between two equally good pages, the better experience tips the balance.
My site has too little traffic for field data. Now what?
Then Search Console shows nothing. Test page by page in the lab and fix what shows red there: the hero image, scripts and image dimensions.
