"Core Web Vitals" sounds like something only a developer needs to understand. It isn't. It's just three numbers Google measures on every website — loading speed, responsiveness, and visual stability — the same way a doctor checks pulse, temperature and blood pressure before saying whether you're healthy. You don't need to know the biology to understand why a high fever is bad news. You don't need to code to understand these three either.
LCP — Largest Contentful Paint
LCP measures how long it takes for the main, largest piece of content on the page — usually a hero image or headline — to actually appear on screen. Think of it as ordering food at a restaurant: it's not about when you walked in, it's about how long you wait before your plate arrives.
According to Google's own current documentation:
- Good: 2.5 seconds or less
- Needs improvement: between 2.5 and 4 seconds
- Poor: more than 4 seconds
On a small-business site, slow LCP is almost always caused by one of three things: an unoptimised, oversized hero image; slow, cheap hosting with a slow first response; or a web font that has to fully download before any text can render.
INP — Interaction to Next Paint
INP measures how quickly the page responds after someone actually does something — taps a menu, clicks "book now", types into a field. It's the elevator-button test: you press it, and the light should react almost instantly. If you press it and stand there for a second wondering if it registered, that's a bad INP.
According to Google's current documentation:
- Good: 200 milliseconds or less
- Needs improvement: between 200 and 500 milliseconds
- Poor: more than 500 milliseconds
On a small-business site, poor INP is usually caused by too much happening at once in the background — heavy booking widgets, chat plugins, tracking scripts and animation effects all competing for the visitor's phone at the exact moment they try to click something.
CLS — Cumulative Layout Shift
CLS measures how much the page unexpectedly jumps around while it loads. Picture reading a newspaper while someone keeps sliding it a few centimetres left and right — you go to tap a headline and a cookie banner has shoved it somewhere else, so you tap the wrong thing.
According to Google's current documentation:
- Good: a score of 0.1 or less
- Needs improvement: between 0.1 and 0.25
- Poor: more than 0.25
On a small-business site, poor CLS is usually caused by images or embedded videos with no fixed size reserved for them, web fonts that render at a different size than the placeholder text, or a cookie banner and pop-up that shove the page content down after it's already loaded.
Why these three specifically?
It helps to notice what each metric is actually standing in for. LCP is a proxy for "does this feel fast to arrive at". INP is a proxy for "does this feel fast to use once I'm here". CLS is a proxy for "does this feel stable and trustworthy while I'm reading it". Together they cover the three moments a visitor can quietly lose patience with a page: waiting for it, interacting with it, and reading it. That's also why fixing only one rarely fixes the overall feel of a site — a page that appears instantly but jumps around every time you scroll still feels broken, just in a different place.
How Google actually measures these
One detail matters more than people expect: Google doesn't grade a page on one lucky fast load. It looks at the 75th percentile of real visits — meaning three out of four real visits to your page need to meet the "good" threshold before Google calls the page good overall, measured separately for mobile and desktop. A page that's fast for you on office wifi can still fail this if it's regularly slow for real visitors on patchy mobile signal, which is exactly the gap that catches most small-business owners off guard: the site felt fine when they checked it themselves.
How to measure your own site for free
You don't need to buy anything to check where you stand, and you don't need to understand the underlying code to read the report.
- PageSpeed Insights (pagespeed.web.dev) — paste in your URL and get an instant score plus a plain-language list of what's slowing you down, with separate tabs for mobile and desktop. This is "lab data": one simulated test run right now, on Google's servers, useful for diagnosing a specific problem.
- Google Search Console — under the Core Web Vitals report (free, just needs your site verified once), you'll see real visitor data called "field data", grouped by Good, Needs Improvement and Poor, based on people who actually visited your site over roughly the last 28 days. This is the number Google's ranking systems actually look at, not the lab test.
The two can disagree, and that's normal, not a bug. Lab data is one simulated snapshot on a clean connection; field data is real people, on real phones, on real networks, at real speeds — including the person standing outside your shop on patchy 4G, who is exactly the visitor these metrics exist to protect. If PageSpeed Insights shows "good" but Search Console shows "poor", trust the field data: it's describing what actually happens, not what's theoretically possible.
Does this really affect your Google ranking?
Yes, but not as a single dominant factor. Google's own Search Central documentation states plainly that Core Web Vitals "align with what our core ranking systems seek to reward" — in other words, a page that scores well tends to already be doing the other things Google rewards, like being usable and not frustrating. Think of Core Web Vitals less as a lever you pull for a ranking boost and more as a floor: bad scores can hold a page back, but great scores alone won't out-rank genuinely weaker content. Good technical performance clears the way for your content and reputation to do the actual work.
What's worth fixing first
If you only have an afternoon, fix in this order — it's ranked by how much improvement you typically get per hour of effort, not by which metric matters most.
- LCP first. It's usually the biggest, cheapest win — compress your hero image (a photo straight off a phone or camera can easily be five to ten times larger than it needs to be for the web), and switch to a modern hosting provider if yours responds slowly. This single change often moves a whole site from "poor" to "good" on its own.
- CLS second. Set explicit width and height on every image and embed, and make sure your cookie banner reserves its own space in the layout instead of sliding in on top of content that's already loaded.
- INP last. It usually needs the most technical work — removing unnecessary scripts and plugins one at a time, and checking whether third-party widgets like chat bubbles or booking calendars are the culprit — so tackle it once the two cheaper wins are already banked.
None of these three numbers is a vanity metric you chase for its own sake. They're a decent proxy for whether a real visitor, on a real phone, is having a smooth experience or a frustrating one on your site — and a frustrated visitor doesn't file a complaint, they just leave.