Working on a website with 5k in 404's - What would you do?

[bwnbwn]

I am working on cleaning up a website and discovered in GWT there are about 5k 404's from one specific content area they had in 2005. For some reason they chose not to pull the data over when they upgraded to a new CMS.

The content is gone deleted from the old DB. All the pages have multiple websites linking to them. This guy was a hot topic back in 2005 and I guess there was something that happened who knows they really won't say.

So with that said what would you do with 5k in 404's that contain hundreds if not thousands on incoming links?

[lucy24]

There are hundreds or thousands of links to content that hasn't existed since 2005? I think it's safe to say that the search engines will know how much those particular links are worth. Change the 404 to a 410, with a well-thought-out 410 page for the rare human visitor, and think no more about it.

[FranticFish]

Hundreds, maybe thousands, of pre-2005 links?

In the absence of any official source to backup the view that those links are:
- definitely worthless right now (I agree, it does sound very reasonable, but Google doesn't always make sense);
- definitely not likely to regain any strength if the url is redirected to new content on a different url;
- definitely not likely to regain any strength if new content is put up at the url...

...I'd say it might be worth trying to work out a way to harness that potential power.

I can't find anything official to indicate what Google's position would be on this. The only slightly relevant thing I could find recent was Matt Cutts saying in 2014 that 404s are ignored for 24 hours in case they're a mistake.

Unless anyone has a direct source or direct experience that would confirm that it should be forgotten, if it were me I'd pick perhaps 5-10 urls with the most links and try to devise a test that would allow me to see if the links do still count, and/or whether there are circumstances under which they could be made to count again.

The potential is there to waste some time, but the potential is there to benefit also.

[RedBar]

I assume you have a customised 404 page that helps people get to where they need/want to be?

You should be able to tell from your own server stats which are bots and which are "real" visitors however, in reality, I'm with lucy24, 410, Gone!

[bwnbwn]

RedBar no they don't yet. I am in the process of putting together recommendations. That currently is one of them.

I am not being paid enough to spend a huge amount of time on this but I did review a lot of the linking domains and there are a large number of them very respectable news sites that I have to assume never run a 404 link checker on their sites, or have so many not worth the time. Either way I was impressed with some of the links.

There is no way to map to an article so the only options really are 410 create a custom 410 page or just leave them a 404 and have them create a good custom 404 page.
410 is the correct response they are gone 10 years ago.

[lucy24]

410 create a custom 410 page or just leave them a 404 and have them create a good custom 404 page.

If you don't want to spend much time, return a 410 response but let the same (physical) page act as an Error Document for both 404 and 410.

[netmeg]

Agree with Lucy.

[bwnbwn]

Done thanks guys