CODEWITHSUNDEEP
debugginggoogle-pagespeedweb-performancecore-web-vitalsinp
John Smith
John Smith

Reputation: 1

INP is over 300 according to Google, but lower than 100 in other tools and real device tests

I am trying to address an Interaction to Next Paint (INP) issue on my webpage. Google's reports indicate that the INP on this page is over 300, but when I test the page using tools like DebugBear or by interacting with it on my smartphone, every click I measure shows an INP lower than 100.

I’ve ensured that the tests include multiple user interactions and have tried different tools to validate the issue, but the INP remains low in these scenarios.

Here are some steps I’ve already taken:

  1. Tested on various devices, including smartphones and desktops.
  2. Used different tools like DebugBear and Chrome DevTools for Lighthouse performance reports.
  3. Simulated various user interactions, including clicks and inputs.

Why might there be a discrepancy between Google’s reported INP and the results I see in real-world and tool-based tests?

Any guidance on identifying and resolving this mismatch would be greatly appreciated!

Upvotes: -1

Views: 39

Answers (1)

Barry Pollard
Barry Pollard

Reputation: 46040

There can be many reasons for why there might be a difference including (but not limited to):

  • Not all interactions are being tested. For example, if there is a login button that pops up a login prompt than are you actually logging in?
  • If this is a Single Page App (SPA) then all INP may be reported at the entry point, even if it happens on a different route.
  • INP interactions in iframes are not available to Web APIs so may be under-reported in tooling. These will be included in Google's CrUX data.
  • Check you're looking at URL-level data and not original-level data which may be reporting for other pages.
  • Users may be on slower devices than what you're testing on.

If you can provide your site, I might be able to role out some of those (or suggest more!) reasons.

But really the best way to narrow this down is by collecting the INP data (ideally with diagnostic attribution details) from your users to see if you can 1) see the same thing and 2) get more details as to why this is. See this article for more information: https://web.dev/articles/find-slow-interactions-in-the-field

Upvotes: -1

Related Questions