Render delay LCP

[Tanya Tritelnitskaya]

Hello!

We face the issue when quite often we have very high LCP render delay. We've already delay our scripts and optimized styles by removing unused CSS. I'd very grateful if you could help me to understand why we are still having that issue as I'm struggling to get that. 
Here is the link https://existek.com/blog/sdlc-models/, but of course you can choose any from our website.

Thank you very much in advance for your time and help!

[Sham]

Hi Tanya, 

For further details, please visit https://web.dev/articles/optimize-lcp.

https://existek3-838c.kxcdn.com/wp-content/plugins/cleantalk-spam-protect/js/apbct-public-bundle.min.js?ver=6.28
 
 https://existek3-838c.kxcdn.com/wp-content/themes/existek/fonts/Roboto-Light.woff2

I am seeing above 2 files indicates render blocking. A quick solution could be to address the font issue by optimizing two files and subsetting the fonts to include only those necessary for above-the-fold content.

[Tanya Tritelnitskaya]

Hello, Sham!

Thank you very much for your reply and recommendation!

I still have more questions, because obviously I misunderstand something. On this page LCP is a text and for the page we actually have the only one font, that is Roboto-Light. Could you please tell me what else can I do with this to prevent render delay? In which other way can we use it?

Thanks a lot in advance!

Bye

[Barry Pollard]

So I presume you are taking the render delay issue from Lighthouse's advice?

https://pagespeed.web.dev/analysis/https-existek-com-blog-hospital-managment-system/3gi79a6kw2?form_factor=mobile

Lighthouse tries to show what might be the issue under certain circumstances, but that may not reflect the actual issue your real users are experiencing, depending on whether they are on similar conditions (network, CPU) to the Lighthouse run. Luckily, you have some CrUX field data which is much more interesting:

This shows you have a TTFB of 2.2 seconds, and FCP of 3 seconds, and then the LCP also happens at 3 seconds (so LCP is not any slower than the first contentful paint).

This tells me that TTFB is your problem, not the render delay, and the Lighthouse audit is not a good explanation here.

With a TTFB of 2.2 seconds you're definitely going to struggle to get LCP under 2.5 seconds and the fact you get 3 seconds (0.8 seconds after the HTML starts arriving) is actually pretty impressive!

To improve your TTFB you could look at our Optimize TTFB guide, but two things I spot immediately are:

• You are using a CDN (keycdn) for your resources, but not your HTML document itself. This has a doubel impact in that i) you're HTML is not benefiting from being globally distributed and ii) it also means the HTML document has to make an extra connection to the keycdn origin before it can download the resources:=. You can see the impact of this delay in the following waterfall diagram, where there are large connection setup requests between 1.0 and 1.5 seconds:

• Your HTML has a cache-control header of max-age=0, meaning it should never be cached.

That means every single request always has to go all the way back to your server every single time. Adding even a short caching time (unless your content does update that much?) and using a CDN for the HTML itself would likely go a long way to solving your issues.

However, your server seems reasonably fast from the above test, so there may be other issues depending how you get traffic to your site. If a large proportion of traffic involves redirects (e.g. from ads, or from newsletter links) then that could also cause TTFB delays. It is worth looking at your traffic sources and seeing if that is the case, and if it is possible to reduce any of those redirects (e.g. if you have multiple link shorteners in use, or are linking to a http site, that then just needs to redirect to https).

Thanks,

Barry

--
You received this message because you are subscribed to the Google Groups "Chrome UX Report (Discussions)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to chrome-ux-repo...@chromium.org.
To view this discussion on the web visit https://groups.google.com/a/chromium.org/d/msgid/chrome-ux-report/bcbff769-c81b-44b2-bbbc-29338a37727cn%40chromium.org.

[Tanya Tritelnitskaya]

Hello, Barry!

I'm very grateful for your fast and detailed reply. Really appreciate all your help!

Best regards!

Bye