Moz-Specific 404 Errors Jumped with URLs that don't exist

[CJolicoeur]

Thanks for the feedback. You're right about the 404 part, I should have phrased it differently. As you figured out, I meant that we are getting 404s for URLs that were never intended to exist and that we don't know how/why they are there.

We are investigating part 1, but my hope is that it is part 2.

Thanks again for taking the time to respond.

[effectdigital]

If you can export the data from Moz and it contains both a link source (the page the link is on) as well as a link target (the created broken URLs) then you might be able to isolate more easily, if it's you or if it's Rogerbot. If Moz UI doesn't give you that data, you'll have to ask if it's at all possible to get it from a staff member, they will likely pick this up and direct you to email (perfectly normal)

[CJolicoeur]

Do you mean downloading the CSV of the issue? I tried that and it gives me the following:

Issue Type,Status,Affected Pages,Issue Grouping Identifier,URL,Referring URL,Redirect Location,Status Code,Page Speed,Title,Meta Description,Page Authority,URL Length,Title Pixel Length,Title Character Count.

Which isn't really useful as it relates to the 404 page.

I'm new to Moz, is there a direct line to an in-house resource that could tell us if it's a Rogerbot issue?

[CJolicoeur]

UPDATE HERE:

I've just noticed that it is also adding the email of the author to the URL and creating an error with that as well.

So, there are three types of errors per post:

OURSITE.com/the-article/COMPANY1

OURSITE.com/the-article/AGENCY1

OURSITE.com/the-article/author@OURSITE.com

[effectdigital]

The ones you want are... URL and Referring URL I believe. "URL" should be the 404 pages, "Referring URL" would be the pages that could potentially be creating your problems

[effectdigital]

No need to freak out though as you say "author@oursite.com" implying they are business emails (not personal emails) so you shouldn't have to worry about a data breach or anything. That is annoying though

[CJolicoeur]

The referring URL in this case is the original url without the added element in the permalink.

So

URL: OURSITE.com/the-article/COMPANY1

Referring URL: OURSITE.com/the-article/

Does that give any more info?

[CJolicoeur]

True, but it's as if something is creating faux URLs of a current article. Adding company names and emails to the end of the URL... It's very weird.

[CJolicoeur]

Other update here..

I've checked about 50 of these errors and they all say the same stats about the problem URL page.

307 words, 22 Page Authority.

I don't know if it matters, just putting it out there.

[effectdigital]

Yeah that will tell you to look on the referring URL, to see if you can track down a malformed link to the error URL look in the coding

[effectdigital]

It is assuredly very weird, we just have to determine if Rogerbot has gone crazy in this Summer heat or if something went wrong with your link architecture somehow

[effectdigital]

At this point that doesn't really matter the main thing is to analyse the referrer URL to see if there genuinely are any hidden malformed links

[CJolicoeur]

I've just come up on something....

In an attempt three days ago to be more transparent (it's a news site), we added "send me an email" links to each author's bio as well as links to the Company and the Agency in the footer.

Except these links weren't inserted correctly in the footer, and half the authors didn't get the right links either.

So instead of it being a "mailto" link, it was just the email which when you hovered over was the url of the page with the author email at the end... the same thing that's happening in the errors.

Same for the footer links. They weren't done correctly and sending users to OURSITE.com/AGENCY1 instead of AGENCY1's website. I've made the changes and put in the correct links. I have asked for a recrawl to see if that changes anything.

[effectdigital]

A more common issue than you might think and strongly likely to be a culprit

[CJolicoeur]

Just wanted to update quickly. The mistakes in the email links as well as the links to the two company sites proved to be the problem. After recrawling the sites, the 7,000+ errors are gone.

It's interesting because I was about to get very upset with Moz, thinking their bot had caused me half a day of headaches for nothing. Turned out they picked up an error before any other system did that would likely have done a lot of damage given that they were all contact links meant to improve transparency.

Hopefully, we caught and fixed the problem in time. In any case, thanks for your help effectdigital.

[effectdigital]

Not a problem! It't great that Moz's crawler picked up on this issue as it could have caused some problems over time, if it were allowed to get out of control