Slow LCP Times for One Folder Only

I’m pleased with way my site appears to performs, but Google Search Console has complained about the same issue a number of times - the mobile LCP time for a number of pages.

These pages look OK to me, and score fine with PageSpeed Insights and web.dev, but the curious thing is that it’s always the same ten pages, and they’re all in the same folder. I don’t believe that it’s anything on the pages themselves as they’re pretty-similar to all of the other pages on the site, but it seems a strange coincidence that of the 300 indexed pages the ten that cause problems are all in the same folder.

Is there any possible reason that performance would vary between folders?

Diagnostic Center | Check SSL and Test Website Security | Cloudflare

May I ask what kind of content do they contain?
Mobile - as always - complains about it due to some JavaScript (I guess) :smiley:

Well, Google. I see this kind of behaviour/reports too from time to time.

Does the Google have to load all the resources on the “folder/directory” page, or rather just list the “indexes”?

From this one:

I see it load fine and fast.

Really, no issue for the end-user.

Just static text, mostly - some have the odd image. Here’s an example page if you want to have a look:

https://www.advanced-ict.info/programming/functions.html

They’re for high-school computing students.

Thanks for your response.

You’re right - I don’t think it’s a problem. I’m just curious about why it’s picking those pages in particular - and always those same ones. It being a Google quirk is probably the answer!

I think there actually is a problem, which causes Google to rate your pages sometimes worse then what they actually are. BTW: thanks for using static sites and building pages that scales and perform good… thanks!

Here is what I think causes the problem:

  • your “Image Sprites” are loaded AFTER the CSS is loaded and parsed.
  • while this itself is not a Problem, the Image Sprites itself seems to be the problem.

Looking at your Google Report it seems unreal how you site does even take longer then 1s to load. But yet Google is reporting LCP (Largest Contentful Paint) of more then 2.5s which is strange.

But after looking how your Image Sprite loads globaly it makes more sense:

In some regions it triggers a 403 Error and even does have loading timings of (this one image itself) of more then 2.5s (NY = 6.4s). Seems like somthing is not correct with this image or how it gets served. I personally would look into this first to find out why Google marks your pages as having a high FCP.

NOTE: this test was done after some hard-reloads and it also depends on from where Google crawls your page. I personally would be happy about a more detailed log/info from Google, but thats how it is.

1 Like

Google Search Console is based on real world Chrome User Experience reports (CRuX) while Pagespeed Insights has both real world field data + synthetic test data. You ultimately want to rely on GSC real world users. See my write up on my forums at https://community.centminmod.com/threads/google-page-speed-insights-and-google-core-web-vital-metrics.20735/

Also linked is LCP outline at https://web.dev/lcp/ which also lists ways to improve LCP and the ultimate way is to improve Time To First Byte (TTFB) and that seems to be where you may fail as on one test for PageSpeed Insights for the URL you posted I get warning for slow server response time (TTFB) but on subsequent tests it was fine.

image

That variance you can check via Google Analytics logging of your average server response times and breakdown that by country/region to see if there’s any specific traffic patterns/sources that are slower pulling down your TTFB averages etc as well as Cloudflare Speed/browser analytics if you’re on CF Pro or higher plan.

From Performance Test - Check URL Speed From 10 Global Locations | KeyCDN Tools

2 test runs

Cloudflare cache certain static content https://support.cloudflare.com/hc/en-us/articles/200172516-Which-file-extensions-does-Cloudflare-cache-for-static-content- but not dynamic/static generated html itself by default (which is what WPT TTFB is testing for). But you can tell Cloudflare to cache dynamic/static generated html content to some extent depending on Cloudflare plan you’re on via cache everything page rule but have to be careful to only do this for static html content and not dynamic html content (otherwise you would cache private logged in user content).

Real world LCP is subject to many factors outside of just the connection itself. I’ve used Elements Timing API to track Core Web Vital metrics in real time on my web sites and even watching a video at the same time as loading the web site in your browser can blow out your timings - example at Performance Tutorials - Google PageSpeed & Webpagetest.org - #4 by eva2000

For forum index page measured H1 title bar rendered times for me in Brisbane, Australia hitting my forums in US West Coast behind Cloudflare

  • Logged in member/visitor no CF full HTML page cache between 2000-2500ms so ~2-2.5 seconds
  • Logged in member/visitor no CF full HTML page cache but watching a video on my computer blew out the H1 tag render time to 19790ms ! Yes 19.79 seconds !
  • Guest logged out visitor with CF full HTML page cache between 900-1300ms with random variation spikes of 2400-4900ms (could be factors local to my PC/apps in use)

This goes to show in real world, your visitors maybe also multi-tasking on their computers which may impact their browsing page load experience and thus impact your real world Google Search Console Speed LCP/FID metrics and Google Page Speed Insight’s reported field/origin metrics :wink:

Below is Opera browser dev tools console reported Elements Timing API render times for my forum index’s title bar H1 tag.

  • left = logged in member = 2096ms render time
  • middle = logged in member while watching a video at same time = 19790ms render time
  • right = guest logged out visitor = 1032ms render time

Given the site is for high school students, more likely they’re multi-tasking on their computers while loading your site i.e. videos etc?

Sounds like comes back to variance in TTFB which effectively impacts all metrics from FCP to LCP.

1 Like

Thanks all for the benefit of your expertise. This is most interesting - especially what’s going on with the image loading time in Singapore. My target audience is the UK (England, actually, as Wales and Scotland have their own curriculum), and most of my overseas traffic comes from the US and India, and those places look better.

I will definitely look into image loading times. The strange thing is that all pages use the same navigation menu, so it still seem strange that GSC only complains about those in the programming folder, but maybe those pages are just more popular in places where loading is slow?

Also, the navigation icons aren’t used in the mobile version. I found a site that would give me a waterfall for the mobile version, and it did confirm that nav_icons.png isn’t loaded, but favicon.ico took about two seconds to load.

I hadn’t really considered that results would be affected by people doing other things at the same time. In practice, my pages are designed to be used on desktop computers in classrooms with fast fibre connections, but I know that Google likes to prioritise mobile results these days.

Plenty of food for thought - thanks.

This topic was automatically closed 15 days after the last reply. New replies are no longer allowed.