https://johnnyreilly.com/2023/01/15/how-i-ruined-my-seo
So, what do you guys think? Maybe it's an update to his back link profit with Google accounting for those 404s?
Okay, I don't play an SEO on the internet, but I'll play along here.
My first thought was that it was just a reporting problem in GA from screwing up the tag, but the huge drop in impressions reported by Search Console shows that it's legit.
>> 404s?
If it's true that a huge number of his pages went 404, then that seems like the obvious reason especially if the backlinks to those 404 pages get disocunted.
Still, I don't follow his whole thing about 404s costing him backlinks. The AHREFs report shows an increase in referring domains. Why would he see an increase in referring domains? I understand that he means it costs him backlinks in Google's eyes, not actual backlinks on the web. But let's say he doesn't lose any backlinks, why does he get an almost 50% increase in referring domains in a couple of weeks on what he says is a very old and stable blog?
It seems like something else odd is happening.
I thought maybe widget backlinks as a result of his mistake posting his GA code (like if it somehow included a link back to his domain), but that is just as described - a basic code snippet for implementing GA4 in Docusaurus, with no link back to his domain. In any case, John Mueller says that these links would just get ignored.
https://developers.google.com/search/blog/2016/09/a-reminder-about-widget-links
https://www.seroundtable.com/google-widget-links-33055.html
Still, something must explain the bump in referring domains and that something might look to Google like a link-building scheme of some sort. Or am I just misreading it? He seems to think the chart shows substantial loss in backlinks, but I don't see that from the charts he posts that supposedly show it.
And regardless, AHREFs does not know what Google does or does not count for backlinks, so that's just an AHREF's crawl. It doesn't really say how many links Google finds and how many Google likes/counts.