Significant differences between CLS in PageSpeed Insights and Chrome DevTools

[littleAmused]

FAQ

• Yes, my issue is not about variability or throttling.
• Yes, my issue is not about a specific accessibility audit (file with axe-core instead).
• Yes, my issue is not answered by other FAQs.

URL

https://www.orchideen-wichmann.de/de/

What happened?

PageSpeed Insights CLS: 0.807
Chrome DevTools Lighthouse CLS: 0.008

https://pagespeed.web.dev/analysis/https-www-orchideen-wichmann-de-de/n8hd7g4gh9?form_factor=desktop

What did you expect?

PageSpeed Insights CLS: 0.807
Chrome DevTools Lighthouse CLS: 0.807

I expected a similar result in Chrome DevTools so that the problem could be debugged more easily.

What have you tried?

Critical CSS generation with help of npm package penthouse

How were you running Lighthouse?

PageSpeed Insights AND Chrome DevTools

Lighthouse Version

latest, I guess

Chrome Version

No response

Node Version

No response

OS

No response

Relevant log output

[arsmediabd]

When you run a Lighthouse test from Chrome DevTools, it is considered a local test. If your website is hosted on a local or very close server, the Lighthouse score will usually appear much higher because of low latency and faster resource loading.

However, Google’s official Lighthouse test reflects global performance. This means it measures how your website performs for users worldwide, not just from your local machine. That’s why the global Lighthouse score can be different (often lower) compared to your local test.

To improve your global page speed score, you should use a CDN—preferably one with an edge network—so your website’s static assets load faster from different geographic locations.

[tunetheweb]

When I clear storage and (Application panel->Storage->Clear site data) and load the page I see significant CLS as measured by the DevTools Performance panel. Reloading it, without clearing storage shows much less CLS.

As Lighthouse does a "cold load" that explains this.