What to do with old content (making the best after Panda)

[dataguy]

Four years and two months ago I had a website which was raking in a cool half a million a year on user created content. Then there was Panda, and after about 6 months I had to take a real job and work for someone else.

I've kept my old website alive with the latest technology (I am a programmer, after all) but it's only in the past few months that traffic stopped going down month after month.

Now I'm wondering what I can do to eliminate the dead weight. My site has so many articles that I have to have a dedicated server to handle it. If I could eliminate the articles which haven't had any readers in, say, 6 months, I could get rid of 70% of the articles and switch to cheaper hosting.

Or course I'm also curious as to the SEO benefits of this. Can I figure out which articles Google deems Panda-hostile by looking at traffic patterns? Can I salvage anything out of this several-hundred-dollars-a-month website?

I'd love to read your advice.

[fathom]

PANDA's focus is about originality.

Having articles that regurgitate any topic is the problem.

I would first eliminate all articles that appear to be derivatives of other works or "spun" in some way. Having a single article about the merits of SEO is devalued by even one other article of similar topics which is common to website that aggregate free articles - so long as the copy was free you were happy to exploit its value... Which there was none and why you ended up needing another job.

In theory, your problem stems from free content which isn't worth much due to that fact. If it was valuable users would not have left it.

[dataguy]

I understand what Panda affects and why my site was dinged. My problem is that while this website (in round numbers) may have around a million worthless articles it still has few hundred thousand unique, compelling articles. There is no way to manually separate the two, despite my best efforts.

My question is if I can take traffic data to figure out which articles Panda deems worthless, and which ones are of some value.

[fathom]

There is no answer to your question.

You will get better results with a few hundred articles than a few thousand or a few hundred thousand.

Addressing the recovery strategy by reviewing the aftermath of PANDA traffic isn't a useful activity.

If you have a few hundred thousand compelling articles out of a milion articles you are suggesting PANDA didn't hit you.

The more likely scenario is, you have a few hundred compelling articles up to a thousand or two.

Pay for the forum membership so you can discuss this in greater detail in the backroom.

[tangor]

One fellow I worked with recently had a similar problem. Our solution was to find 100 good, traffic keeping articles and put all the rest behind a subscription wall (the price was email address and name), then locked the SEs away from that content. All that was left, the 100 forward facing, was the cream of the crop AND daily insert of new content with an eye to that being the best possible. Within four months traffic was up 80% and engagement was up by 20%. Surprisingly, the subscription side showed a decent membership. This move preserved the old material, but kept it out of the SEs ... the users find it all by themselves. :

[fathom]

Excellent suggestion tangor!

The catalyst isn't the more content the better results or more traffic it's the least amount of content to produce the most returns or the top 1% or less of content should be crawlable not some value based on minimal traffic levels.

I was able to do a search on your username to locate your linkedin profile and now know more about your domain in question... Tangor suggest is a valid way to achieve recovery.

[netmeg]

Yea, that sounds to me like it would def be worth a shot.

[trabis]

I am in a similar situation. Would 'noindex' 90% of the ugc be a solution? i.e. keep only indexed the most upvoated or viewed(refered) content?

[fathom]

The point isn't about percentage as if PANDA nails you your total website was likely garbage.

That said, deindexing the bulk will do the trick.

[dataguy]

Yes, Tangor, that is advice I can use.

So the next question is, do I noindex the pages, return a 403 Not Authorized for requests which aren't from logged in users, return a 410 or a 404? The last time I noindex'ed pages, google kept them in their index anyway, but this was several years ago. I know that with 404'ing a page, the Googlebot may continue trying to retrieve it for years.

[aakk9999]

I also really like Tangor's suggestion.

With regards to the response code, I would return 403 Not Authorised with an inviting logon/register page.

[tangor]

As a followup, we used 403.

There was no concern for link juice as the site had been trashed (specifically by G) and after making these changes the site continued to function and is gaining lost ground.

That said, the site had to change as some of it was clearly crap and the UGC had been unmonitored for too long.

This only works if the site has a reason to live and the webmaster has a dedication to bring things right. It is not a magic bullet.