- This topic has 8 replies, 3 voices, and was last updated 3 months, 3 weeks ago by David.
September 16, 2020 at 3:39 am #1445421David
my blogposts have to high LCP values (4,6s).
How can this be reduced to a normal value, so that Google likes it?
This post is good for analysis:
This is, what Google PageSpeed tells me, about the Element
Largest Contentful Paint-Element 1 element found
Dies ist das größte Inhaltselement, das im Darstellungsbereich angezeigt wird. Weitere Informationen
Flipchart gestalten – die WIRKLICH wichtigen Tipps für deinen Erfolg 1
It´s the featured image and has only 63kb. But takes 4,6 seconds to load on mobiles. On desktop the value lies at 1,2s.
What can be done to reduce it?
I´m using WordPress, GeneratePress + WP Rocket.
DavidSeptember 16, 2020 at 4:30 am #1445487DavidStaffCustomer Support
it looks like you’re featured image is still lazy loading – on your previous topic, the code Leo provided may not of worked as the HTML for the
<img>is being changed by whatever method your using to output WebP images.
Can you disable the WebP images – so we can test without it ?September 16, 2020 at 8:20 am #1445943David
Hi David, i used imagify to create and present WebP´s.
Now it´s deactivated, but the LCP is still high at 3,6s.September 16, 2020 at 9:12 am #1446028David
Could it be helpfull, to preload the images above the fold?September 16, 2020 at 9:43 am #1446073DavidStaffCustomer Support
The bigger issue is the FCP time – any improvements made to that will improve the LCP.
I ran a GT Metrix report from a London based server. Report is in the Private information.
And the results are quite impressive.
Check out the waterfall report.
Couple of immediate fixes that will remove 2 x requests can be completed by changing the Customizer > General > Icon Type to SVG.
The other issues are the green ( waiting to connect ) and purple ( requesting ) lines seen in the waterfall report. Thats something your host may have an idea on – BUT considering it has a 0.9sec load time i don’t think they’ll be too bothered.
The two major ones are the first two .css files that have to be loaded before the page can be rendered. As your servers request type is HTTP 2.0 ( Which can handle a lot more concurrent server requests ) you may ( or may not ) see an improvement if you stop Combining CSS in your Cache plugin. That way it can load a lot of little files at the same time as opposed to a couple of large files.
Now the fun part – re-running the test on a US server and the times are a lot longer – like 4.4secs total load. So i recekon the LCP time we are seeing is a false negative, as the tests are being completed on a remote google server. Over time as more localised users access the site you should see an improvement on the Field Data in Pagespeed Insights.
If of course you’re targeting a global audience you may want to talk to your host about setting up a CDN.September 17, 2020 at 3:43 am #1446874David
Good ideas, David.
But they didn´t worked.
Switching Icons to SVG didn´t made any progress.
And disabling the combine function resulted in much more request and worse performance.
I have uploaded my whole testing sheet under:
DavidSeptember 17, 2020 at 2:45 pm #1447803TomLead DeveloperLead Developer
Here’s the report David mentioned: https://gtmetrix.com/reports/sinnstiften.biz/CS0mfg5q – let us know if you want it removed once viewed.
I’m not sure there’s much left there on the theme-side of things to tweak. It seems like the biggest issue you’re having is with server response time in the US (as David mentioned above).
As a test, do things improve significantly if you remove the featured image?September 25, 2020 at 4:00 am #1458257David
for testing i deactivated the featured image.
FCP: 2.2 to 1.7
LCP: 3.6 to 2.5
It´s faster, but has not that big effect.
Of course, whithout the image, is very ugly now.
DavidSeptember 25, 2020 at 7:51 am #1458652DavidStaffCustomer Support
You can run the test locally by opening the post and Right Click > Inspect to open Chrome Dev Tools. From there select and run a Lighthouse report.
You will see the lab data it provides is significantly better then the results Goole is returning. This demonstrates that the remove servers Google is using is taking a lot longer to complete its requests.
Over time, if your users are Local to your server, the Field Data will get updated to reflect the real response times.
Please note that LCP and the other new metrics are not actually being used in googles ranking algorithim, it is scheduled to do so in 2021.
- You must be logged in to reply to this topic.