Core Web Vitals tools compared: field data, lab data and quick checks

Field data is a 28-day distribution of real loads, so it is the number to judge yourself by — but it lags, and a low-traffic page may not have enough samples to report at all.

Lab audits run instantly and tell you why a load was slow, which is what you need while fixing. Use lab output to choose a change and field data to confirm it worked weeks later.

On content sites the largest element is the hero image or the first block of text waiting on a webfont. Compressing that image properly, serving it at display size, and giving fonts a fallback removes most of the delay.

Compress before upload rather than relying on the server: a browser-based compressor keeps the file local and lets you compare quality settings side by side.

Layout shift is caused by elements without reserved space — images missing width and height, late-loading banners, embedded widgets. Reserve the box and the shift disappears.

Interaction delay is caused by long tasks on the main thread, almost always third-party scripts. Removing one analytics or chat widget usually beats every micro-optimisation you could make to your own code.

Pick your five most important URLs. Run a lab audit on each, note LCP and the largest asset, fix the single worst item, then re-check. Repeat next month rather than trying to perfect everything now.

Track it beside your rank and crawl checks so performance work stays proportional to the traffic it protects.