What Are Core Web Vitals?

Core Web Vitals are a subset of the web performance metrics Google treats as fundamental to a page's user experience. Google announced them in May 2020, and they now sit among the technical SEO signals Google groups under "page experience," alongside HTTPS, the absence of intrusive interstitials, and solid mobile rendering.

Today there are three: LCP measures perceived loading speed, INP measures responsiveness to interactions, and CLS measures visual stability. They don't replace the rest of the page speed indicators or a full performance audit; Google picked these three as a ranking signal because they apply to every kind of site, not just one narrow category of page.

The trio has changed once since launch: in March 2024, INP replaced First Input Delay (FID) as the third metric, after a year and a half as an experimental one. Any definition still listing FID as an active part of Core Web Vitals describes the state before that switch.

These thresholds apply to the 75th percentile of a page's visits, calculated separately for mobile and desktop: a page misses the "good" cut if a quarter of its visits perform poorly, even when most of them load fine.

INP is the newest of the three. It officially replaced FID as a Core Web Vital on March 12, 2024, after running as an experimental metric since May 2022. The practical difference is that FID only measured the delay of the first interaction, while INP scores the page's responsiveness across the entire visit, catching interactivity problems FID never saw.

Core Web Vitals get measured two fundamentally different ways, and mixing them up is the most common mistake when reading the numbers. Field data comes from the Chrome User Experience Report (CrUX), collected from real visits by Chrome users who've opted in; it reflects how a page actually loads, with each visitor's own connection and device. Lab data, by contrast, comes from tools like Lighthouse, which simulate a visit under fixed conditions. It's useful for catching a page speed problem before it ships, but it doesn't reflect real usage.

The Core Web Vitals report in Search Console is built on field data, not on Lighthouse's simulations: it measures what real visitors experienced on a given URL, or on a group of structurally similar URLs, aggregated by week. When a page doesn't get enough traffic to have its own CrUX data, Google falls back on the numbers from other pages with a similar structure on the same origin.

That's why a URL can score high in a lab test and still show up as "needs improvement" in Search Console: they're two separate measurements, taken at different times under different conditions, and only the field data reflects what the people who actually visited that page went through.

Core Web Vitals have been a ranking signal since 2021, folded into the wider set of page experience signals Google uses in search. A good score doesn't guarantee the top spot: content relevance to the query still carries more weight. But between two equally relevant pages, the faster, less frustrating one has an edge.

What a business feels first is visitor behavior. Slow loading or an interface that lags behind a click pushes people to abandon a purchase, a form, or an article before they finish it. That drop-off happens regardless of search position, so optimizing these metrics pays off in conversion even if the ranking never moved an inch.

That's why it makes more sense to treat them as part of the site's UX rather than a box to tick for the search engine. A slow LCP on a product page, or a CLS that shifts the buy button right as someone taps it, costs sales before it costs rankings.

• Prioritize the above the fold element that's likely your LCP, such as a header image or a large headline, and don't let third-party scripts with no visible payoff delay it.
• Cut back on JavaScript that runs during user interaction: breaking long tasks into smaller chunks improves INP without touching the design.
• Reserve space with width and height (or aspect-ratio) for images, videos, and iframes, and avoid dropping a banner or cookie notice on top of content that already loaded.
• Serve images in modern formats like WebP, with lazy loading outside the initial viewport, without delaying the ones that actually make up the LCP.
• Check the Core Web Vitals report in Search Console by URL group, not just a single page's score in a lab tool.
• Test on real mid-range devices, not just a developer's laptop: Google indexes and scores mainly the mobile version under its mobile-first approach, and that's usually where the gap shows up most.

• Looking only at the lab score from a tool like PageSpeed Insights and never checking the real field data that actually drives the Search Console report.
• Optimizing LCP with a compressed image while still loading fonts or scripts that block rendering before it appears.
• Mixing up INP with FID and still talking about "first input delay" as the current metric; FID was retired on March 12, 2024.
• Fixing CLS by ripping out animations the interface actually needs, instead of simply reserving the space those animations will occupy.
• Treating Core Web Vitals as a one-off project: without ongoing monitoring, a new deployment can quietly wreck an INP that had been green for months, and nobody notices until the next technical SEO review.