Core Web Vitals in 2026: what you need to know.
Google’s performance metrics have tightened. Here’s what to measure, what to actually fix, and why lab scores can mislead you.
Core Web Vitals became a ranking signal in 2021. Since then the thresholds have tightened, the metrics have evolved, and the proportion of sites that actually score well has stayed frustratingly low. If your site is slow or unstable and you know it, this is a practical starting point for what to actually address.
The Three NumbersThe three metrics
that matter.
Core Web Vitals currently centre on three measurements. Largest Contentful Paint (LCP) measures how long it takes for the main visible content of a page to render. Interaction to Next Paint (INP) measures how quickly the page responds to user input. Cumulative Layout Shift (CLS) measures how much the page visually jumps while it’s loading.
The passing thresholds are LCP under 2.5 seconds, INP under 200 milliseconds, and CLS under 0.1. Most sites that fail do so on LCP and CLS. The causes are consistent: unoptimised hero images, render-blocking resources loaded in the document head, and content that shifts because dimensions aren’t specified on images and embeds.
LCPLCP is almost always
about images.
If your LCP score is poor, the problem is almost certainly the hero image or the largest above-the-fold element. The fixes are well-established: serve images in WebP or AVIF format, use correct dimensions, avoid lazy-loading the LCP element (a surprisingly common mistake), add a fetchpriority="high" attribute so the browser prioritises it in the load queue, and serve it from a fast CDN close to your audience.
For sites on Shopify, WordPress, or similar platforms, image handling is often where the most gains come from with the least engineering effort. Theme image helpers and media optimisation plugins can change scores significantly without custom development work.
CLSCLS is almost always
about missing dimensions.
Layout shift happens when the browser doesn’t know how large something is before it loads. An image without specified width and height attributes, an embed without a defined aspect ratio, a font that loads late and shifts surrounding text, a cookie banner that drops in after the initial render. These are the most common culprits.
The fix for images is adding explicit width and height attributes in the HTML, which allows the browser to reserve space before the file arrives. For fonts, using font-display: swap and preloading critical font files reduces the magnitude and delay of font-related shift. For third-party widgets, a fixed-height wrapper that reserves space regardless of when the content loads prevents the jump entirely.
A slow page doesn’t feel slow. It feels untrustworthy.
Urbanframe Studio, Web & Digital 2026
Measure field data,
not lab scores.
The most important step in improving Core Web Vitals is measuring field data (real user measurements), not just lab scores from Lighthouse. Google Search Console provides this through its Core Web Vitals report. PageSpeed Insights shows both lab and field data for specific URLs. These are different things and they need to be read differently.
Lab scores from Lighthouse are useful for identifying what to fix. Field data is what affects your rankings and your users. A site that scores 90 in PageSpeed lab testing but has a slow 75th-percentile LCP in field data has a real-world performance problem that the lab score isn’t representing. Fix what the users are actually experiencing. That’s the number that matters.
