Website performance work starts with real-user experience, not one test score. This guide explains LCP, INP, and CLS and prioritises fixes on pages that affect traffic and revenue.
Quick summary
What you get from this article
- Use field data to decide and lab data to diagnose.
- Good targets are LCP ≤2.5s, INP ≤200ms, and CLS ≤0.1 at p75.
- Hero images and JavaScript are often the highest-value starting points.
- Core Web Vitals support experience and Search but cannot compensate for weak relevance.
- Prioritise pages that generate traffic and revenue.
Separate real-user data from test scores
Field data reflects real visitors across devices and networks over time. Lab data simulates one visit in a controlled environment and is better for reproducing causes before release.
If Lighthouse looks healthy but Search Console does not, investigate device mix, page groups, and third-party scripts that differ from the test, then allow field data to accumulate after the fix.
Image slot prepared — artwork to follow
Comparison of real-user field data and a controlled one-run lab test
Read LCP, INP, and CLS correctly
Slow LCP means the main content arrives late, often because of the hero image, server response, or render blocking. High INP means the page responds slowly after interaction, often because JavaScript monopolises the main thread. High CLS means content shifts because space was not reserved.
- Good LCP — 2.5 seconds or less
- Good INP — 200 milliseconds or less
- Good CLS — 0.1 or less
- Evaluate the 75th percentile separately for mobile and desktop
Image slot prepared — artwork to follow
Visual explanation of LCP, INP, and CLS using events on a web page
A high-value order of fixes
Start with pages that already generate organic traffic or conversions instead of changing every template at once. Record the before state, field baseline, and post-release outcome.
- Compress and right-size images
- Preload only critical above-the-fold images or fonts
- Remove external scripts and split non-critical JavaScript
- Improve caching, CDN use, and server response
- Reserve dimensions for images, embeds, banners, and consent UI
Image slot prepared — artwork to follow
Website speed priority ladder from images and fonts through JavaScript, caching, CDN, and layout stability
How speed compares with other SEO work
Google says Core Web Vitals are used by ranking systems, but there is no single page-experience score that guarantees position. When two pages are similarly relevant, better experience can contribute; a fast irrelevant page still loses.
Once priority pages pass, better content, internal links, and conversion work usually beat pushing a lab score from 95 to 100.
Image slot prepared — artwork to follow
Matrix comparing content relevance with page experience to prioritise SEO work
Preparation
Information to prepare before you start
- PageSpeed Insights for priority pages on mobile and desktop
- The LCP element and long-running scripts
- Image, font, and JavaScript weight
- Field-data baseline before changes
- Conversions for each priority page
Related services
