Core Web Vitals explained without the jargon

Largest Contentful Paint measures when the biggest visible thing on the screen finishes loading. Usually that is a hero image or a headline. It is the closest single number to "when did the page become useful".

Google treats under 2.5 seconds as good and over 4 seconds as poor. For context, a large share of visitors abandon a page well before four seconds, so a poor LCP costs you people before it costs you ranking.

The usual culprit is an oversized image. The usual fix takes an afternoon.

Cumulative Layout Shift measures how much content jumps around as the page loads. You have experienced this: you go to tap a link, an advert loads above it, and you tap something else entirely.

Good is under 0.1. Poor is above 0.25. It is expressed as a unitless score, which is unhelpful, but the practical meaning is simply how much the page moved.

Almost always caused by images without dimensions, or by content injected above what the visitor is already reading.

Interaction to Next Paint measures the delay between tapping something and the page visibly reacting. It replaced an older metric called First Input Delay in 2024 because that one only measured the first interaction, which flattered most sites.

Under 200 milliseconds is good. Above 500 is poor. High values usually mean too much JavaScript running on the main thread, so the page looks ready while being unable to respond.

This is the one people misdiagnose as "my site is slow" when the page actually loaded quickly — it simply ignored them afterwards.

Core Web Vitals are part of how Google ranks pages. They are not the largest factor — relevance and content quality matter far more — but they are a genuine tiebreaker between pages that are otherwise comparable.

The more useful framing is commercial rather than technical. These three numbers describe whether your page is pleasant to use on a phone. Google measuring it is a reason to care sooner, not the reason to care.

A speed test runs a simulation on a machine somewhere. That is lab data: repeatable, useful for diagnosis, and not necessarily what your actual visitors experience.

Field data comes from real Chrome users visiting your site. It is the truth, but only exists if your site has enough traffic for Google to collect it. A new site will show lab data only, and that is fine — treat it as diagnosis rather than verdict.

Fix LCP first. It is the most commonly poor, the most directly tied to people leaving, and usually the easiest win because it is nearly always an image.

Then CLS, which is tedious but mechanical. Leave INP for last: it is the hardest to fix and typically requires changing how the site is built rather than adjusting settings.