The Core Web Vitals are three metrics from Google measuring a page's user experience: how quickly the main content appears (LCP), how much the layout shifts while loading (CLS), and how quickly the page responds to clicks (INP). You can check these values for free with PageSpeed Insights or in the Search Console report. The most common cause of poor scores on SME websites is oversized images, and you can fix that yourself.
Anyone wanting to improve their website's load time runs into three abbreviations: LCP, CLS and INP. These aren't programming terms, they're three questions every visitor asks themselves without realising it. Does the content appear quickly? Does the page stay still while loading? Does it respond when I click something? Google measures these three questions and calls the answers Core Web Vitals.
The three metrics in plain language
| Metric | The question behind it | Good (as of October 2026, according to web.dev) | Typical cause on SME websites |
|---|---|---|---|
| LCP (Largest Contentful Paint) | When is the largest element in the visible area loaded, usually the header image or the main heading? | up to 2.5 seconds | Oversized header images, slow hosting, too many fonts |
| CLS (Cumulative Layout Shift) | How much does the content shift while the page is loading? | up to 0.1 | Images without fixed dimensions, late-loading banners or cookie notices |
| INP (Interaction to Next Paint) | How long does it take for the page to visibly respond to a click or input? | up to 200 milliseconds | Too many scripts from plugins, chat widgets, embedded maps |
Google distinguishes three levels: good, needs improvement and poor. According to Google's documentation, what counts for ranking is whether a page reaches the "good" threshold. Optimising LCP from 2.4 to 2.5 seconds therefore has less impact than improving from 5 seconds to 2.5.
A real-world example: a physiotherapy practice has a photo of the team on its homepage, uploaded straight from the camera at 6 MB. On a mobile network, this image alone takes several seconds to load. The LCP is poor, even though the rest of the website is lean. Shrink the same image to 200 KB, and the score moves into the green.
Where to measure the metrics
You don't need any software or server access for this.
- PageSpeed Insights (pagespeed.web.dev): enter a page's address and wait for the report. At the top are field data from real visitors, if Google has enough of it. Below that is the lab measurement with a list of specific recommendations, such as "serve images in modern formats", along with file names.
- Search Console, "Core Web Vitals" report: shows how many URLs across your whole website score good, needs improvement or poor, split by mobile and desktop. Setup is described under Setting up Search Console.
- Chrome on your phone: open your website with Wi-Fi switched off. This doesn't replace proper measurement, but it shows what your customers actually experience.
Don't just measure the homepage. For many SMEs, the most-visited page is a service page or a blog post.
What you can fix yourself
Most poor scores on small websites come down to a handful of causes that can be fixed without any coding knowledge.
- Shrink images. Before uploading, scale them to the width at which they're displayed (1,200 to 1,600 pixels is usually enough for a content column), and save them as WebP or compressed JPEG. The rules for this are under Image SEO. This is the lever with the biggest impact on LCP.
- Set image dimensions. If your system doesn't automatically add an image's width and height, the browser won't reserve space for it, and the text will jump as soon as the image loads. Current WordPress themes handle this automatically; with older themes, the setting in the image block helps.
- Clear out plugins and widgets. Every chat button, every social media widget, every embedded map loads its own scripts. Ask yourself of each element: has it brought in a customer in the last three months? If not, remove it.
- Don't auto-load videos. An embedded YouTube video on the homepage loads several hundred kilobytes of scripts before anyone even clicks play. A preview image with a link, or a plugin that only loads on click, solves this.
- Reduce fonts. Two font styles are enough for most websites. Every additional one delays the text.
What's out of your hands
Honestly, some problems aren't something a business owner can fix themselves. If PageSpeed Insights shows a high "time to first byte" value, the hosting is usually too slow. In that case, the only fix is switching provider or plan. If a theme or website builder loads too much code by default, often the only option is a different theme. And if the layout jumps because of a cookie banner, the banner provider needs to fix it.
In these cases, it's worth sending the PageSpeed report with the specific error message to your agency or host, rather than simply reporting that "the website is slow". The report names the affected files, which significantly shortens the discussion.
The order that works
Improve the page with the most visitors first, not the one with the worst score. Shrink the images, remove unnecessary embeds, then measure again. If the scores are still poor after that, it's down to hosting or the theme, and that's something to discuss with a specialist. Core Web Vitals aren't a project you finish once, but they don't need monthly attention either. A check once a quarter, and after any major redesign, is enough for most SME websites. What SEO covers beyond the technical side is explained in What is SEO?.
Frequently asked questions
How much do Core Web Vitals affect my ranking?
Google names Core Web Vitals as part of the signals for user experience, but weights content more heavily. A page with poor scores can still rank well if it provides the best answer. Between two pages with similar content, the faster one can have the advantage. For SMEs, that means content first, then loading speed, but don't ignore either.
Why does PageSpeed Insights show different figures than Search Console?
PageSpeed Insights measures a single page in the lab and also shows field data from real visitors, if enough is available. Search Console groups pages by similar issues and relies solely on field data. Small websites with few visitors often have no field data at all, in which case the Search Console report stays empty.
Is a caching plugin enough to improve the figures?
A caching plugin speeds up the delivery of the page from the server and often helps with LCP. However, it doesn't fix problems rooted in the content, such as a 4 MB header image or a layout shift caused by a banner loading late. Clean up images and embeds first, then activate caching.
Should I measure the figures on mobile or on a computer?
On mobile. Google primarily evaluates websites in their mobile version, and the figures there are almost always worse, because the connection and processor are slower. PageSpeed Insights shows both versions, but the mobile one is what counts for decisions.