[Support request] Long LCP on Desktop & Mobile

[John]

Hi guys – I’m on Siteground cloud hosted, but despite my best efforts, still seeing long LCP times for both mobile and desktop. Any ideas how to fix this?

Links shared in private section. Thanks!

[Fernando]

Hi John,

I can see that Siteground is already optimizing your scripts which is part of the cause of the issue.

Perhaps you can also try using a CDN? Using a CDN should improve delivery of your site’s resources.

A popular CDN would be Cloudflare. This link might also be helpful: https://www.wpbeginner.com/showcase/best-wordpress-cdn-services/

[John]

Actually I’m using the SG CDN as well. I used to use Cloudflare but the SG CDN is supposed to work better with their servers because everything is under one roof. Any other ideas? Or if you need any other info, please let me know. Thanks.

[Fernando]

Can Siteground defer Scripts? If so, can you try that?

Alternatively, here are some plugins that you can try:

https://wordpress.org/plugins/async-javascript/
https://wordpress.org/plugins/flying-scripts/

[John]

Yes, all the JS that can be deferred is. Flying scripts seemed to create weird double-tap issues on mobile (links would need a double tap to be triggered). But yeah the scripts are already deferred.

[David]

Hi there,

you could make some improvements regarding images.
1. Query Loop blocks, if you select the GB Image Block, you can set the Attachment size to a smaller option.
2. Lazy loading – i see some images are being lazy loaded but others are not, both in query loops and in single post content.

That will reduce image size and defer the off screen images. Which should make an improvement.

LCP
on a pagespeed report the top data is Real User data that has FCPs of around 2.8 and LCP of 3.3.
This is aggregated data collected across the domain over the last 28 days.
Below that is the lab diagnostic data, which is for that specific test. The FCP and LCP are 1.6 and 1.9 respectively. Which is great.
Over the next 28 days you should start to see the Real user data improve to reflect the lab data.

[John]

Thanks David! I actually had already set the query blocks to attachment size medium large, and the lazy loading inconsistencies seem to be an issue with SGs lazy loader skipping some images (they’re skipping the ones in the sidebar and the footer of the posts, and the query blocks on the homepage) which they hopefully will fix in an update soon.

In terms of reducing unused CSS and JS, I’m using an asset unloader, which I assume happens before SG combines and minifies the CSS and JS.

Yes I have to wait for the next 28 data to kick in. For some reason, even after I had these good numbers, the last 28 day collection still failed even on desktop, which is surprising. But sometimes the specific test data is great, and other times the LCP is slow.

If you have any other suggestions or see anything that stands out, please let me know. Thanks.

[David]

Make sure you have page caching enabled on the server.
And ask SG to do some checks on the server. I ran a few different tests using dev tools and webpagetest.org on one of your single posts and the TTFB time varies quite greatly from 0.5s to almost 2s. Its very inconsistent which could be something else eating up server time.

[Author]

Posts