Core Web Vitals in 2026: the three numbers that decide whether you rank

Google scores how your pages feel to real visitors using three measurements, known together as Core Web Vitals. Each one has a pass mark, and you need all three.

Largest Contentful Paint, or LCP, is how long the biggest thing on screen takes to appear. That is usually your headline or your hero image. The pass mark is 2.5 seconds or less.

Interaction to Next Paint, or INP, is how quickly the page reacts when somebody taps, clicks or types. The pass mark is 200 milliseconds or less. INP replaced the older First Input Delay measurement and is far stricter: instead of watching only the very first tap, it looks at every interaction during the visit and reports the slowest one.

Cumulative Layout Shift, or CLS, measures how much the page jumps around while it loads. This is the text that slides down just as you go to tap a link. The pass mark is 0.1 or less.

Across the web, roughly 43 percent of sites still miss the 200 millisecond INP target. Layout shift is the easiest of the three to fix and has the best pass rate. Largest contentful paint sits in the middle. INP has the worst, because it is not a plugin problem.

Slow interactions almost always trace back to too much JavaScript running on the visitor's device. Chat widgets, tracking scripts, sliders and third-party embeds all compete for the same processor, and the most expensive one wins the race against your visitor's tap.

Tools such as PageSpeed Insights show two sets of numbers: a lab result from a simulated run, and field data collected from real people over the past 28 days. Google uses the field data.

That distinction matters. A perfect lab score is a diagnosis, not proof. If your field data still shows a poor INP, real visitors are still waiting. Always read the field numbers first and treat the lab run as a way to reproduce the problem.

You do not need a rewrite. Most failing sites are fixed by the same short list of changes:

An AI assistant has to fetch and render a page before it can quote it. Slow, script-heavy pages are harder to read and get cited less often, while sites that show something almost immediately turn up far more frequently in AI answers.

So the work you do for Core Web Vitals now pays twice: once in search rankings, and again in whether an AI assistant names you when somebody asks for a recommendation.

Start with real field data rather than opinions about the design. Fix the template that covers the most pages first, usually your homepage or your main service page, because a slow template drags down every page built from it.

Then re-measure after 28 days, which is the window the field data uses. Change one thing at a time so you can tell what actually worked.