Core Web Vitals

What it is

Three numbers, each with a threshold Google calls “good”:

• LCP (Largest Contentful Paint) — when the biggest thing in the viewport finishes rendering. Good is 2.5 seconds or less. On most pages the biggest thing is the hero image or the headline.
• INP (Interaction to Next Paint) — how long the page takes to paint a response after a tap or a click. Good is 200 milliseconds or less. It replaced FID in March 2024 and is much harder to pass, because it measures every interaction, not just the first.
• CLS (Cumulative Layout Shift) — how much the layout jumps around while loading. Good is 0.1 or less, on a unitless scale. Late-loading webfonts, images without dimensions and injected banners are the usual causes.

The scores come from real Chrome users over a rolling 28 days, and the pass mark is the 75th percentile — so the slowest quarter of your visits are allowed to be bad.

Why it matters if you are paying for a website

Less than the people selling audits imply, and more than the sceptics say. Google has been clear that page experience is a tiebreaker rather than a lever: it will not push a thin page above a better one. What it does do is decide between pages that are otherwise equally good answers — which, for a local business competing with four similar sites, is most of the market.

The stronger argument has nothing to do with ranking. LCP is the number that correlates with people leaving before the page appears, and CLS is the number that correlates with mis-taps. Both are conversion problems that happen to be measurable.

What I actually see on projects

Two patterns, over and over.

The first: a Lighthouse score of 100 next to failing field data. Lighthouse runs one page, once, on a simulated connection, with no cookie banner and no third-party tags — which is not the visit your customers make. The Search Console report is the one that counts, because it is the only one built from real sessions. That gap is what my SEO monitoring dashboard was built to watch, and it opens on almost every WordPress site with an analytics stack bolted on.

The second: on migrations off WordPress, LCP improves mostly by deletion. Not through image tuning but by removing render-blocking work — the plugin CSS, the font loader, the slider script — and shipping HTML that was finished before the request arrived. INP improves for the same reason: there is no main-thread work left to block the response.

What Core Web Vitals are not

They are not a speed test, and they are not a target you hit once. A page that passes on the day of launch can fail three months later because marketing added a chat widget. They also say nothing about whether the page is any good: a fast page that answers the wrong question still loses.