CLS et CRux !

[Claude BATAILLE]

 Why does the CLS of CRux vs. Lighthouse diverge?

[❄ Johannes Henkel]

CrUX publishes field data, which is collected from real users. So the CLS is not  an individual measurement but a statistical aggregation.

Lighthouse simulates a page visit, so it runs a localized test of a specific page load. This is also known as lab data. It's useful for diagnostics, e.g. sometimes it may give you an idea how to improve the CLS, but it doesn't always align with the statistical distribution from the field.

On Mon, Jun 1, 2026 at 1:47 PM Claude BATAILLE <bcomb...@gmail.com> wrote:

 Why does the CLS of CRux vs. Lighthouse diverge?

--
You received this message because you are subscribed to the Google Groups "Chrome UX Report (Discussions)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chrome-ux-repo...@chromium.org.
To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ux-report/facb9bd5-d01f-4e35-bad9-8ece63a2872cn%40chromium.org.

[Barry Pollard]

We also give a lot of advice here: https://web.dev/articles/optimize-cls

Two key points that site owners often don't consider are:

1. Lighthouse measures initial load CLS, whereas CrUX includes all CLS on the page — including post load CLS as the user scrolls, or otherwise interacts with the page.
2. Lighthouse is often used to test one page (often the home page), whereas CrUX can show CLS on individual pages (if there is enough data) and the whole origin. There is a toggle in the top right of PageSpeed Insights allowing you to switch between the two. If you have no CLS on the home page, but CLS on other pages, then that usually indicates you are not testing the right pages. Google Search Console attempts to group pages to help identify page types which have the issue so is often a good place to look.

Thanks,

Barry