How do you improve Core Web Vitals?
Work from field data grouped by URL pattern rather than a single lab score, because the problem is usually a template rather than the page you tested. Then fix Largest Contentful Paint first — the hero image and server response — since it is the metric that most commonly fails.
Measure real users, not a lab test
A page-speed score from a single test run is a diagnostic, not a measurement. What counts is field data — real visits, on real devices, on real connections.
Search Console's page experience reporting shows field data grouped by URL pattern, which tells you which page types have a problem. Use the lab tool afterwards to work out why.
The common trap is optimising the homepage because that is what you tested, when the problem is on a template used by two hundred service pages.
Largest Contentful Paint
This is time until the biggest visible element renders — usually a hero image or a headline block. Almost always the metric that fails.
What fixes it, in order of typical impact:
- Resize and compress the hero image; serve AVIF or WebP
- Do not lazy-load it — preload it instead
- Remove render-blocking CSS and JavaScript from the head
- Inline the small amount of CSS needed for the initial view
- Fix slow server response — caching, database queries, cheap hosting
- Self-host fonts, or preconnect and use
font-display: swap
Server response time is the one people skip and it sets the floor for everything else. If the server takes 800ms to send the first byte, no front-end work gets you under that.
Interaction to Next Paint
This measures how long the page takes to respond visibly when someone taps or clicks. It is a JavaScript problem almost every time.
- Remove scripts you do not need — old tracking, unused libraries, abandoned plugins
- Defer everything non-essential
- Break long tasks so the main thread is not blocked
- Audit third-party tags; chat widgets, heatmaps and tag managers are frequent offenders
On a typical small business site, deleting third-party scripts nobody looks at is the single most effective change available.
Cumulative Layout Shift
Things moving while the page loads. Causes and fixes:
- Images without width and height — always set them
- Ads or embeds without reserved space — reserve it in CSS
- Web fonts swapping in at a different size — size-adjust the fallback
- Content injected above existing content — banners, notices; render them in place
This one is usually the cheapest to fix and is frequently ignored because it does not show up in a static screenshot.
Work on templates, not pages
A fix applied to one page helps one page. The same fix in the template that generates two hundred pages helps all of them. Group your problem URLs by template before starting.
Keep expectations proportionate
Core Web Vitals are a real factor but a modest one compared with relevance and content quality. A fast page about nothing does not rank.
The better argument is commercial: a slow page loses visitors before they read anything, and on mobile connections that loss is substantial. Speed pays for itself in conversions well before it pays in rankings.
Frequently asked questions
Should I trust a page-speed score?
As a diagnostic only. What counts is field data from real visits on real devices. Search Console groups it by URL pattern, which tells you which page types have a problem — often a template, not the homepage you tested.
What fixes Largest Contentful Paint?
Resizing and compressing the hero image, preloading rather than lazy-loading it, removing render-blocking CSS and JavaScript, and fixing server response time — which sets the floor for everything else.
What causes poor Interaction to Next Paint?
JavaScript, almost always. Removing unused scripts, deferring non-essential ones, breaking up long tasks, and auditing third-party tags like chat widgets and heatmaps are the effective fixes.
How do I fix layout shift?
Set width and height on every image, reserve space in CSS for ads and embeds, size-adjust fallback fonts so text does not resize on swap, and render banners in place rather than injecting them above content.
How much do Core Web Vitals affect rankings?
They are a real but modest factor next to relevance and content quality — a fast page about nothing still does not rank. The stronger argument is commercial: slow pages lose visitors before they read anything.