Core Web Vitals: What Still Matters After the 2026 Changes

Google’s Core Web Vitals changed again in 2026, and most teams reacted the same way: they opened PageSpeed Insights, chased the lowest score and shipped a dozen micro-optimisations. A few weeks later the field data barely moved. This guide is the order of work we use on real client sites, where the numbers that count come from real visitors, not lab runs.

What actually changed in 2026

The three metrics are still Largest Contentful Paint (LCP), Interaction to Next Paint (INP) and Cumulative Layout Shift (CLS). What changed is how strictly the thresholds are applied on mobile, and how much weight INP now carries for pages that rely on heavy client-side JavaScript.

• LCP stays at 2.5 seconds for a “good” rating, measured at the 75th percentile of real visits.
• INP is the metric most sites now fail, usually because of long tasks from third-party scripts.
• CLS is mostly solved on modern themes, but ads, embeds and cookie banners still break it.

None of this is new in principle. The difference is that a page which scraped through last year on desktop numbers alone can now fall into “needs improvement” on mobile, and that is where most of your traffic is.

Start with field data, not lab scores

Lab tools are great for debugging, but they run on one device over one network. Search uses the 75th percentile of real visits over the last 28 days. Before you change a single line of code, open the Core Web Vitals report in Search Console and group the failing URLs by template: product pages, blog posts, category pages and so on.

Fix templates, not pages. One change to the product template can move ten thousand URLs at once.

Which report to trust

Use the Chrome UX Report for origin-level trends, Search Console for groups of URLs, and your own real-user monitoring script when you need per-interaction detail for INP. When the three disagree, trust the one with the most real visits behind it.

Set a baseline you can compare against

Write down the current 75th-percentile LCP, INP and CLS for each template before you start. Field data takes up to 28 days to fully reflect a change, so without a baseline you will not know which fix moved the numbers.

The fix order that moves the numbers

We work through the list below in this order, because each step makes the next one easier to measure. Skipping ahead to fine-grained tweaks while a chat widget blocks the main thread for 400 milliseconds is the most common way teams waste a sprint.

1. Remove or delay third-party scripts that block the main thread.
2. Serve the LCP image early: preload it, size it, and never lazy-load anything above the fold.
3. Break long JavaScript tasks into smaller chunks so the browser can respond to input.
4. Reserve space for anything that loads late, such as embeds, ads and banners.

Images and fonts

Most LCP problems are image problems. Use modern formats such as WebP or AVIF, give every image explicit width and height, and keep the hero image out of any slider. Fonts matter too: self-host them, preload the one used above the fold and use font-display: swap so text appears straight away.

JavaScript and INP

INP measures the slowest interaction a visitor makes on the page. A single heavy click handler can fail the whole page, so profile the interactions people actually use: menus, filters, search and add-to-cart. Long tasks can be split so the browser gets a chance to paint between them:

async function processItems(items) {
  for (const item of items) {
    render(item);
    // Give the browser a chance to handle input between chunks.
    await scheduler.yield();
  }
}

Where scheduler.yield() is not available, a setTimeout of zero does a similar job. The point is the same: never hold the main thread for more than about 50 milliseconds at a time.

WordPress-specific wins

On WordPress sites the biggest gains usually come from the plugin list, not the theme. Every plugin that loads scripts on every page adds to INP, even when its feature is only used on one page.

• Load form, slider and gallery scripts only on the pages that use them.
• Use page caching and a CDN so the HTML itself arrives quickly.
• Replace page-builder sections that ship large unused CSS with lighter templates where it matters most.

How to keep the gains

Performance regresses quietly. A new marketing tag, a heavier hero video or an extra plugin can undo months of work in a single release. Add a performance budget to your CI pipeline, track field data every month and review each new third-party script before it ships.

A simple monthly check

Once a month, compare each template’s field data with your baseline, list any new scripts added since the last check and confirm the LCP image on your top pages is still the one you expect. Ten minutes a month is far cheaper than a rescue project.

Key takeaways

Measure with field data, fix at the template level, tackle third-party scripts first and protect the results with a budget. That order is what turns a red report green and keeps it there. If you would like a second pair of eyes on your site, our team can run a free Core Web Vitals review and send you a prioritised fix list.