> It's not one thing that cause the drop but a combination of things

Let me propose a bold generalization based on observations of sites of all sizes wrestling with Google over the last 20 years, and to which there are certainly exceptions (this case may be one):

It’s always one thing.

Of course, there are always a multitude of improvements that can be made to nibble around the edges and get incremental improvements in various metrics. And a gradual change in traffic may have several simultaneous sources.

But when the effect is an identifiable precipitous drop in search traffic, almost all the time, it turns out there was a single reason, whether a change Google made or a change the site made.

This may be a totally unhelpful observation, and feel free to ignore it. But in these situations I’ve come to find it’s more fruitful to look for A Cause than to approach it as “a little here, a little there.”

I've listed all the causes I can think of. Does any strike you as "the one"?

Honestly, the thing about other sites using your GA tag seems like the biggest red flag/potential for some Google algorithm to have actively started penalizing you. The timing doesn’t quite track, but maybe some interaction between that and the spam update?

> Add a robots.txt

Likely not the cause but not having a robots.txt can cause weird crawling problems.

Since it takes two minutes to set up, why not. And, here's a tip, don't try and be clever, keep it simple:

  User-agent: \*
  Disallow:

Thanks for the tips!

> A cool trick to improve the result fast is by removing the lazy load effect from the LCP: https://i.imgur.com/rOOWm91.png

Haha, ironically I'm responsible for wholesale lazy loading of images in Docusaurus:

https://github.com/facebook/docusaurus/pull/6598

Generally that's a good thing, but it's not good for title images that appear "above the fold". I wonder if there's something we could do to determine whether an image is "above the fold" and so not apply lazy. Something to ponder.

> Try to make more strategic articles stronger.

Not sure I follow your meaning here, do you literally mean applying bold to certain articles? Or something else?

That's cool. Just for fun, try to remove the lazy load effect from the LCP, let's see how it effect the score.

As for the strategic articles, I mean that you'll increase the internal links certain articles you think can rank higher.

On October you had 3300+ keyword that were ranked 11-100 on Google

https://i.imgur.com/ZGU0Ezk.png

It dropped to 165 keyword now. That's an insane drop, but I wonder how much traffic these thousands of keywords brought you since there weren't rank too high.

I would like to dive into the keyword that ranked 4-10, on Google (50 keywords). And they keywords that ranked 1-3 (15 keywords)

By understanding on GA which pages brought you the most traffic/ranked the highest you can make them stronger by adding more internal links to these pages, look at it as kind of first aid ;)

EDIT:

Few tips:

- The 'Recent posts' widget is great. It get your article indexed faster. What about adding a widget below for 'Popular articles'? Place 5-10 articles there.

https://i.imgur.com/MBquX6M.jpg

- Make sure you implement time stamp to your articles the right way

https://i.imgur.com/b5usnJJ.png

https://i.imgur.com/0vpXdyE.png

tell me more about the timestamp. Where does that data get derived? I thought that should be in great shape since I set lastmod in my sitemap based on git commit date

https://johnnyreilly.com/2022/11/25/adding-lastmod-to-sitema...

We could apply lazy loading only after the 2nd image maybe

Yeah - primitive but probably reasonable. Would be happy to experiment with that. I've also been pondering things we could do around open graph images as well, would be nice to use image CDNs like Cloudinary for open graph images in the same way we can for blog images https://johnnyreilly.com/2022/12/26/docusaurus-image-cloudin...

Changing all images to be lazy loaded is not a good idea for SEO reasons (and perhaps not a good idea for regular usage of the site either). If it were, browsers would just do it automatically. Using hints on your images is only useful if you are doing it strategically, which means only lazy loading images that are offscreen on the initial page load. Otherwise you are not actually giving useful hints to the browser on how to load your page in the correct manner, and thus you are just better off letting the browser use whatever internal logic it has to decide how and when to load the assets.

Basically you can think about the optimal, minimal set of resources to render your page being:

- HTML

- Required CSS

- Above the fold images

Then at that point, download all of the other things to make the page work in the way you want (javascript, etc). Anything else is delaying the initial page-load. Because your images are lazy loaded, your page load looks like the following:

- HTML

- Required CSS

- Some above the fold images (profile.jpg, etc)

- Massive, >300kb blobs of javascript (runtime-main.js, main.js)

- Above the fold images in the post

This is not good. It doesn't make sense for your page to download 300kb of javascript before downloading the 18kb image that is above the fold. Now you can partially solve this problem by making the javascript asynchronous, but that still is just another band-aid on the problem, as then the javascript and above the fold images would download concurrently, which is still not optimal.

What you want to do is have above the fold post images be loaded eagerly (the default), and then lazy load ones that are lower on the page. If you aren't going to do that, you probably are better off just not having the images being lazy loaded at all, especially if your page includes 300kb of javascript which is likely going to be much larger than the combined size of all the images on the page.

Yeah I'd love to get your help! My email address is johnny_reilly at hotmail dot com or you can DM me here: https://twitter.com/johnny_reilly or here https://fosstodon.org/@johnny_reilly

Edit II:

Few more useful tips:

- The drop might be related to the Google algorithm update. This link can help with understanding if the traffic drop is in correlation with your traffic drop:

https://ahrefs.com/google-algorithm-updates#october-2022-spa...

- Google search console: Check if there was any penalty More info here:

https://searchengineland.com/google-penalties-manual-actions...

EDIT III: This thread might be useful

https://support.google.com/webmasters/thread/185308786/sudde...

More related threads:

https://www.google.com/search?q=site%3Ahttps%3A%2F%2Fsupport...

> You don't have to get 100 score, but passing the core web vitals score and having higher score on mobile is recommended

Note that they don't have a CWV score yet due to low traffic. But a 39 performance score from the simulated Lighthouse is often more than enough for a passing grade. That is: if a Moto G4 can do OK, your normal users will likely do great.

For instance, a site I made[0] has a 22 from Lighthouse, but a passing CWV grade, so further improvement to the LCP, FID, and CLS would confer no direct Google SEO benefit.[1] (But it may help things like bounce rate, which may confer second-order benefits)

> by removing the lazy load effect from the LCP

Indeed. Even better, making it high priority instead of normal: https://addyosmani.com/blog/fetch-priority/

[0] https://i.imgur.com/TGD1sj2.png

[1] "For example, a page with an LCP of 1750 ms (better than the “good” LCP guidance) and another one with 2500 ms (at the “good” guidance) would not be distinguished on the basis of the LCP signal" – https://support.google.com/webmasters/thread/104436075/core-...

This is very helpful, thank you. Raised a ticket against Docusaurus to track this: https://github.com/facebook/docusaurus/issues/8552 will experiment