Check explainer · v1
Core Web Vitals in the lab: LCP, CLS and main-thread cost
Last reviewed
Three lab grades cover why the page feels slow: what painted last, what moved, and how much work the main thread and the DOM carry. They come from one render, not from field data.
How this check works
Three grades run on the rendered page. Largest contentful paint reads the largest image or text block and warns above 2,500 ms, keeping the element summary, the resource URL and FCP/TTFB as reference. Cumulative layout shift sums the load-time shift score and warns above 0.1; a score of exactly 0 is a measured pass. Main-thread cost warns when total blocking time is over 200 ms, the DOM holds more than 1,500 elements, or the tree is deeper than 32 levels. Every value is a lab measurement from the local render lane — not a CrUX, PageSpeed Insights or Search Console field value.
What does a failure look like?
A 2,600 ms LCP with a hero image that arrives after the first paint; a CLS of 0.18 from a banner inserted above the article; a page with 180 ms of blocking time from a synchronous analytics bundle plus a 2,100-element DOM — each is a warning, not a failed page.
How to fix it, step by step
- For LCP, compress and preload the resource named in the finding, remove head scripts that block the first paint, and shorten the server response.
- For CLS, reserve width and height or aspect-ratio on images and embeds, avoid inserting content above what is already painted, and animate with transform.
- For main-thread cost, split and defer non-critical JavaScript, virtualise long lists, and keep the DOM under 1,500 elements and 32 levels deep.
What does a pass or fail look like?
Good
LCP 1,840 ms, CLS 0.02, total blocking time 90 ms and a 900-element DOM — three passes from one render.
Fail
LCP 3,200 ms, CLS 0.16 and 410 ms of blocking time — the page loads, but every first view pays for it.
Frequently asked questions
Is this my CrUX score?
No. CrUX is a field distribution of real visitors; this is one lab render on the audit's network and device. Use the lab number to find the cause and CrUX to judge the population — they answer different questions.
Why does CLS 0 count as pass while a missing LCP counts as na?
A CLS of 0 means the render was measured and nothing moved. A missing LCP entry means the browser never reported a candidate, so there is no signal to judge. The audit keeps those two cases apart instead of turning both into a pass or both into a warning.
Does a warning hurt rankings?
No. These are warning-level lab signals measured against community thresholds, not a Google ranking system. The audit also does not turn them into a Lighthouse score or declare Core Web Vitals passed.
This rule is one of the checks the page scan runs. Run the free page scan
Free preview; no account needed.