Forum Moderators: Robert Charlton & goodroi

Message Too Old, No Replies

Core Web Vitals A Lot Worse On Real World Data

         

mattlimbert

10:42 am on Aug 7, 2023 (gmt 0)



Hi all,

Wondering if people have come across this issue before, we've had a client come to us confused about their Page Speed Insights score - It's a lot worse on real-world data compared to the simulated environment which we've never seen.

For context:
Sim:
  1. LCP @ 1.4s
  2. CLS @ 0.001

Real World:
  1. LCP @ 4.6s
  2. CLS @ 0.16


Stumped here. The real world TTFB is 3.7s but when running through tools like GTMetrix and doing tests internally we're looking at sub 100ms.
The site isn't slow so we're not sure what's affecting the scores. Has anyone had any experience with issues like this?

Thanks,
Matt

engine

11:09 am on Aug 7, 2023 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



Yes, and it could be the hosting.
Is it hosted on professional servers?
What does a simple test such as ping, or traceroute indicate?

mattlimbert

11:58 am on Aug 7, 2023 (gmt 0)



Hi engine,

Yes it's hosted professionally, there are other websites on the server that don't experience the same issue, which is one of the reasons we're so confused.
The ping results show an average of 8.51ms response time.
It's almost like the data that Google's got for Page Speed Insights is coming from users with really poor internet connection...

Thanks,
Matt

engine

1:54 pm on Aug 7, 2023 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



Matt, what about the possibility that you're seeing cached data (which will seem much better, and that the problem is with the site?
Also, consider that testing from a different location will show different results: For example, if hosted in the U.S, and tested from Europe, it may show different results if hosted in U.S. and tested in U.S. Thant's one of the reasons why I suggested a Traceroute test.

With any site test you should consider that the site might just be a bit slow. For example, a database driven site might be slower or faster than a smaller static site. Or it could be graphics that need optimising, etc.

There are other test tools out there to try.

Another test to try, if practical, is to try hosting the site temporarily elsewhere. That's not always practical (site size and complexity), and a last resort, imho.

Nutterum

2:17 pm on Aug 8, 2023 (gmt 0)

10+ Year Member Top Contributors Of The Month



This smells like testing from another region. Keep in mind that if you test from your country you will have better results, but google may test from where your hosting provider is, which may as well be another country , resulting in worse outcome.

JeffChang

10:47 am on Aug 24, 2023 (gmt 0)



Use ChromeUX reports if you don't already. You'll get a lot more real world valuable data from actual users.

Setup a LookerStudio report and get digging