To high LCP on mobile devices

Hi there,

It’s possible to move the featured image, but likely not out of the first viewport.

What you could do is hide the featured image on single posts, then manually add it to your content out of the initial viewport.

Are you sure the image is to blame? Are your scores better if you temporarily remove it?

Hello Tom, in my tests always the image is marked as LCP:

<img width="1200" height="800" src="https://bzvqew.myraidbox.de/wp-content/uploads/2021/01/Gartenblog_Topfgarte…" alt="Aussaatkalender Januar - 7 Gemüse, Blumen und Kr?uter auss?en - Gartenblog…" />

And how I do hide the image and manually add it again? Isn’t there a possibility to make the header-element larger (in the pro-version) so that the image is moved down automatically? For example with margin-bottom: xyz px?

https://webpagetest.org/result/210221_DiGG_e238144f742c818ec51a6965a73f64a1/1/details/#waterfall_view_step1

Greetings Kathrin

• This reply was modified 3 years, 9 months ago by Steven Stern (sterndata).

Hi there,

To help out:

I think what Tom meant was you can remove the featured image on the post page by using this PHP snippet:

add_action( 'wp', function() {
    remove_action( 'generate_before_content', 'generate_featured_page_header_inside_single', 10 );
} );

and then re-add the featured image as an image block somewhere within the content that isn’t visible to the initial viewport.

Thank you very much! I will try it!

Greetings Kathrin

No problem. Let us know how it goes.

For the moment I have left the Featured Image where it was in the first viewport. The blog is now online with GeneratePress and I get so many different values for the LCP, which one is correct?

Mobil:
Largest Contentful Paint 3,2 s (PSI)
Largest Contentful Paint 1.055s (WebPageTest.org, Server Location Frankfurt)
Largest Contentful Paint 2.4 s (Lighthouse)

Desktop:
Largest Contentful Paint 1,1 s (PSI)
Largest Contentful Paint 2.4 s (Lighthouse)
Largest Contentful Paint 0.712s (WebPageTest.org, Server Location Frankfurt)
Largest Contentful Paint 0.9s (GTMetrix, Server Location Vancouver)

Tested Link was: https://www.topfgartenwelt.com/aussaatkalender-januar-7-gemuese-blumen-kraeuter

Which result is the right one?

Greetings Kathrin

Hi there,

There is no right or wrong result here.

Each testing server will generate different results, depending on their location, how the network is throttled, what testing method they use etc..

PSI and Lighthouse both use Lantern algorithm, so for mobile devices particularly its a simulated result. Not a real time result.

Whereas GTMetrix and WebPageTest use actual devices based on various servers around the world.

Presumably your server is closer to Frankfurt as it generates the best results on WebPageTest.

The real LCP value will be collected in Google Field Data – this will come from every user that visits your site. Each of their individual LCP times will differ depending on their location, the network they are on and the device they are using.

If your site visitors are predominantly located in Germany and using good networks and modern devices then you should see timings similar to the WebPageTest results.

Thank you David for the answer. So I have to wait and hope that things are getting better. At the moment the LCP for Desktop is 6,5s in Search Console ??

Well good news is – its a lot better on the PSi – so you should hopefully see the Search Console numbers improve over the next few weeks – you can try recrawling the site to see if bumps them up sooner.

• This reply was modified 3 years, 9 months ago by David.

The problem is, it was never bad on PSI. I had always good scores on PSI also with the old theme, but suddenly too high scores in the Search Console. The LCP was growing within 2 days about 2 seconds and I haven’t changed anything on the website.

Since midnight the blog is online with GeneratePress and I don’t know which scores I should believe in….

Greetings Kathrin

Perhaps this sheds some live on Google PSI vs. Lighthouse discrepancies.
https://stackoverflow.com/a/64792982

Thanks for the info, but the scores in Lighthouse and PSI are always better as in Search Console. In the meantime the LCP in the Search Console has grown to 7,4s for Desktop and PSI measures an LCP for Desktop of 0,8s. I don’t know why the scores are increasing so drastically from day to day in Search Console and the reason therefore can’t be measured anywhere. Because on every measuring tool the scores are normal.

When I run a test in Google Dev Tools in incognito mode I get this results:
desktop: 9700 ms https://imgur.com/a/447EP3z
mobile: more than 8000 ms https://imgur.com/a/jiMZCvv

Greetings Kathrin

• This reply was modified 3 years, 9 months ago by topfgartenwelt.

Google Console will capture any metrics for any site visit – so every time you visit the site it will capture the LCP time. So if you or any of your visitors are running on a slow network then a longer LCP time will be captured…. so you may be affecting the scores by continuously running tests over a throttled network.

Ok, but I’m situated in Austria, my Server is in Germany and I only have readers from Austria, Germany and Switzerland. In all 3 countries there is no problem with slow internet-connections for desktop devices and also not for mobile. The mobile standard is a fast 4G connections and sometimes already 5G.

Greetings Kathrin

Then there should be no issue with LCP…
If your site currently has very few visitors – I would recommend that you stop testing your site on a throttled network as YOU may be negatively effecting your LCP times.

Other than that i cannot explain what the issue is … its not GP/WP related.