Every year someone declares Core Web Vitals dead, and every year the sites that ignore them keep quietly losing customers. In 2026 the stakes are higher, not lower: page experience signals influence both classic rankings and how often AI answer engines cite your content. This is the exact playbook our engineers run before any client site goes live.

Start with the field, not the lab

Lab tools like Lighthouse are useful, but they describe a single device in a data centre. Your users are on a three-year-old Android on a train outside Sheffield. Field data from the Chrome UX Report (CrUX) tells you what real visitors experience at the 75th percentile — and that is the number Google evaluates. We pull field data first, lab-test second, and never confuse the two.

The 2026 thresholds that matter

  • LCP (Largest Contentful Paint): under 2.5 seconds at the 75th percentile. On UK mobile networks this is a build budget, not a hope.

  • INP (Interaction to Next Paint): under 200 ms. Since INP replaced FID, every interactive element is measured — carousels, filters, menus, everything.

  • CLS (Cumulative Layout Shift): under 0.1. Almost always caused by images without dimensions and fonts without fallback metrics.

Performance is not a feature you add at the end. It is a budget you agree at the start — and CI enforces better than any developer's memory.

The five fixes that do most of the work

Across the last forty audits, five issues account for the overwhelming majority of failed vitals: unoptimised hero images, render-blocking third-party scripts, web fonts loaded without size-adjust, layout shifts from late-loading ads or embeds, and main-thread work triggered by every interaction. None require a rewrite. All require someone to own them.

Set a budget in kilobytes and milliseconds, wire it into your deployment pipeline so a regression fails the build, and review field data monthly. That single habit — enforcement over intention — is the difference between a fast site at launch and a fast site in three years.