Core Web Vitals Assessment: Passed, But Mobile performance score is 33 only

[Amit]

I am seeing a webpage's Core Web Vitals Assessment: Passed, BUT Mobile performance score is 33. How it is possible ?
How it passed even First Contentful Paint 4.0 s, Largest Contentful Paint 10.9 s and Total Blocking Time 2,530 ms

See attached image.

[Barry Pollard]

The Core Web Vitals assessment is based on what real users experience (i.e. the top section of that report). In that you can see that LCP is passing, and FCP (which is not a Core Web Vital but an "other noteable metric" is almost passing too:

The data beneath this is from a Lighthouse run and is used to diagnose more information from a lab-based simulation under specific environmental conditions (a slowish device and connection typical of the global average). Ideally these would be roughly equal so the diagnosis information is useful, however it's not uncommon for them to diverge like you see for your site, and there could be many reasons for this, including:

• Your users may typically be visiting on faster devices or connections than the test was run under.
• Your users may be made up of many repeat visitors with cached assets, and therefore load faster than the "cold load" the test is run under.
  
• Your typical users may have handled consent so may or may not be seeing that consent banner shown in the test screenshot.

In these cases, what the real users are experiencing is the actual fact and so is much more important (which is why the assessment uses that).

However, the lab-based data can still be useful to see opportunities, especially for certain classes of users (first time visitors on slower devices/networks). It also looks like your LCP at 2.4 seconds is just under the recommended "good" threshold of 2.5 seconds and, as I mentioned, FCP is not in the "good" category. So it may still be worthwhile to look at the Lighthouse recommendations to see if you can give yourself more headroom here, even though the provided lab LCP (10.9 seconds) and Lighthouse Score (33%) do not appear to be reflective of your actual users. Improving these in the lab is likely to reflect positively on your real user Core Web Vitals numbers—even if you don't get them to all the way 2.5 seconds and 100 as it seems you don't need to based on the fact that your user profile seems to be better than those used by the PSI Lighthouse test.

--
You received this message because you are subscribed to the Google Groups "web-vitals-feedback" group.
To unsubscribe from this group and stop receiving emails from it, send an email to web-vitals-feed...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/9994391a-111b-4a83-ab2a-907e0b209be2n%40googlegroups.com.

[Amit]

Thanks for reply.

My question is in first top heading LCP is showing 2.4 seconds, but in just it's bottom LCP is showing 10.9 seconds.

But it is passing core web vitals. how it is possible. One side LCP showing 2.4 s and seconds LCP is 10.9 s.

Much Big differences, How it is possible ?

[Barry Pollard]

For the reasons explained:

it's not uncommon for them to diverge like you see for your site, and there could be many reasons for this, including:

• Your users may typically be visiting on faster devices or connections than the test was run under.

• Your users may be made up of many repeat visitors with cached assets, and therefore load faster than the "cold load" the test is run under.

• Your typical users may have handled consent so may or may not be seeing that consent banner shown in the test screenshot.

 

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/d9083cff-f039-4daa-9eae-9e0fde99a4fan%40googlegroups.com.

[Amit]

Thank for explain.

But is it due to Fake spoof speed optimization ?

[Barry Pollard]

I don't know what you mean by that.

Are you saying people trying to game the Lighthouse score? Typically they go the other way (i.e. make the score better rather than worse!). It's not great "fake scoring" if it's making it worse :-)

You can try running Lighthouse locally in DevTools (which typically is faster than PSI as developers often have been machines than those used by PSI) to see if that's more representative. From the Lighthouse module you can also control the settings used. PSI uses basically the same global settings for all tests (minus testing from your local region where possible) so as I say, it may not be representative if your global users are significantly better than the global average.

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/8371ee68-6243-47e6-840f-8da449324e69n%40googlegroups.com.

[Amit]

Yes, My means "People trying to game the Lighthouse score. Make the score better rather than worse! They are showing Spoof score.

[Barry Pollard]

But they aren't better than real-life LCP scores. They appear to be considerably worse. So if they are doing that, then they aren't doing a very good job of it!

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/94c6a944-b297-4bc6-9eaa-2fde755baa37n%40googlegroups.com.

[Amit]

Yes, That's why I contacted you here. I am suspecting because 1 side LCP is 2.4 and second side LCP is showing 10.x seconds.

There are something wrong. Look like Fake Spoof speed work done in that website.

[Barry Pollard]

Well I've no reason to believe this is that, rather than the reasons I've given and, as you have not provided the URL it's not possible to investigate this further.

Lighthouse uses Chrome so if there is some "fake spoofing" going on affecting the real Core Web Vitals to make them seem faster than it is, then it's likely to affect Lighthouse too (but doesn't appear to be).

And if there is some "fake spoofing" going on to deliberately make the Lighthouse scores seem slower than they really are, then that is very odd.

The most likely reasons, however, are for the reasons I gave and, as I say, you've given no reason to believe that isn't the case.

To view this discussion on the web visit https://groups.google.com/d/msgid/web-vitals-feedback/c83d4eee-dead-4355-96f1-47a99ab214c8n%40googlegroups.com.

[Tony McCreath]

People selling speed improvement services sometimes fake it for lab tools like Page Speed Insights and Lighthouse. 

It's hard to tell if your site has such code.

A common clue is the presence of highly obfuscated code on the page.

Another clue is if you have paid a developer to speed up your site or have installed a plugin/app to do the same.

[Amit]

Ok.