Most visitors don't leave a slow page because they hate the brand. They leave because waiting feels like friction, and friction on the first screen kills conversion. Core Web Vitals are Google's attempt to measure that friction in a way both developers and site owners can discuss.
The three metrics, translated
LCP, Largest Contentful Paint
How long it takes for the main content of the page to appear. For most sites, this is the hero image, the main heading or the first product photo. Good LCP feels fast; bad LCP feels like the page is still thinking.
INP, Interaction to Next Paint
How responsive the page feels once you try to interact with it, clicking a button, expanding a menu, opening filters. Bad INP feels like the page ignores you for a beat.
CLS, Cumulative Layout Shift
How much content jumps around while the page is loading. Bad CLS is the feeling of reaching for a button and watching it move somewhere else, usually because an ad, image or banner loaded late.
Why these metrics affect business, not only SEO
- Slow first paint means people close the tab before the content appears.
- Unresponsive interactions make forms, filters and checkouts feel broken.
- Layout shifts cause misclicks, which create frustration, returns and support tickets.
- Google uses these metrics as a ranking signal, so poor performance also reduces how often you're found in the first place.
The impact is cumulative: small performance problems add up across sessions, and small improvements compound too.
What usually makes the biggest difference
- Images. Correctly sized, modern formats (WebP/AVIF), lazy-loaded below the fold, and with explicit dimensions so the browser doesn't have to guess.
- Fonts. Limit to what you actually use, self-host when possible, and preload the one that renders the hero text.
- Third-party scripts. Analytics, chat widgets, A/B tools and tag managers are often the biggest source of slow INP. Load them deferred, and only on pages where they're needed.
- Unused JavaScript. Shipping less code means less parsing, less execution, less energy on low-end devices.
- Server response time. A slow backend puts a floor on everything else. Caching, query optimisation and a sensible hosting tier matter.
- Reserved space. Use explicit width/height or CSS aspect ratios for images, ads, iframes and banners so the layout doesn't shift later.
Common mistakes
- Measuring only on your own fast laptop. Users are on mobile networks and mid-range devices. Use field data (real user monitoring) alongside lab tests.
- Chasing a single number. A perfect Lighthouse score on one run doesn't mean much; consistent field metrics across weeks do.
- Adding tools to fix performance. More scripts rarely make a site faster.
- Treating performance as a one-off. Without monitoring, it regresses each time the team adds a new widget.
A sensible sequence
- Get baseline field data for LCP, INP and CLS on your top 5 pages.
- Fix the hero image and main heading first, they disproportionately affect LCP.
- Audit third-party scripts; remove or defer anything not strictly needed.
- Reserve space for images, ads and late-loading elements.
- Monitor continuously, and treat regressions like bugs.
