Best way to handle different views of the same page?

[ChatterBlock]

Say I have a page:  mydomain.com/page

But I also have different views:

/?sort=alpha

/print-version

/?session_ID=2892

etc.  All same content, more or less.

Should the subsequent pages have ROBOTS meta tag with noindex?  Should I use canonical?  Both?

Thanks!

[AlanMosley]

This is a hard one, cannonical is the easy choice, but Bing advises against it, as you should not have a canonical pointing to itself, it could lead to lose of trust in your website. I would not use the robots for this as you lose your flow of link juice

http://www.bing.com/community/site_blogs/b/webmaster/archive/2011/10/06/managing-redirects-301s-302s-and-canonicals.aspx

I would try to no-index follow all pages excpt for the true canonical page using meta tags, this means some sort of server side detection of when to place the tags.

[Dr-Pete]

I'll 80% agree with Alan, although I've found that, in practice, the self-referencing canonical tag is usually fine. It wasn't the original intent, but at worst the search engines ignore it. For something like a session_ID, it can be pretty effective.

I would generally avoid Robots.txt blocking, as Alan said. If you can do a selective META NOINDEX, that's a safer bet here (for all 3 cases). You're unlikely to have inbound links to these versions of your pages, so you don't have to worry too much about link-juice. I just find that Robots.txt can be unpredictable, and if you block tons of pages, the search engines get crabby.

The other option for session_ID is to capture that ID as a cookie or server session, then 301-redirect to the URL with no session_ID. This one gets tricky fast, though, as it depends a lot on your implementation.

Unless you're seeing serious problems (like a Panda smackdown), I'd strongly suggest tackling one at a time, so that you can measure the changes. Large-scale blocking and indexation changes are always tricky, and it's good to keep a close eye on the data. If you try to remove everything at once, you won't know which changes accomplished what (good or bad). It all comes down to risk/reward. If you aren't having trouble and are being proactive, take it one step at a time. If you're having serious problems, you may have to take the plunge all at once.

[AlanMosley]

Yes i agree for most it is probably not going to be a problem, But Duane again yesterday blogged about this, he did say they can live with it. but they dont like it, and the best thing is to fix it. http://www.bing.com/community/site_blogs/b/webmaster/archive/2011/11/29/nine-things-you-need-to-control.aspx

this leaves me in 2 minds, he said that they may lose trust in all your canonicals if they see it over used, this can be a worry if you have used it for its true use elsewhere.

I also worry about lose of link juice, as Duanes words in the first blog post were, "Please pass any value from itself to itself"

does that mean it loses link juice in the process like a normal canonical does?

I myself would fix it anouther way, but this may be a lot of work and bother for some. Thats why I say its a hard one.

[ChatterBlock]

Thanks gentlemen.  I will probably just go with the NOINDEX in the robots meta tag and see how that works.

Interesting side note, SEOmoz will still report this as a duplicate page though ;-(  Hopefully the search engines won't.

[AlanMosley]

I am not sure if SEOMoz will, but search engines wont as it wont be in their index.

[Dr-Pete]

I haven't seen evidence they'll lose trust yet, but it's definitely worth noting. Google started out saying that, too, but then eased up, because they realized it was hard enough to implement canonical tags even close to correctly (without adding new restrictions). I agree that, in a perfect world, it shouldn't just be a Band-aid.

[AlanMosley]

What do you think about what Duane said, about assigning value to itself, could this be a LJ leak as it would be a leak if it was assigning value to anouther page?

[Dr-Pete]

I generally trust Duane, so I'd take it at some value - I just haven't seen that problem pop up much, practically. Theoretically, you'd create a loop - so, if it leaked, it would keep looping/leaking until no juice was left. That seems like an odd way to handle the issue.

My bigger concern would be the idea that, if you rel-canonical every page, Bing might not take your important canonical tags seriously. They've suggested they do this with XML sitemaps, too - if enough of the map is junk, they may ignore the whole thing. Again, I haven't seen any firm evidence of this, but it's worth keeping your eyes open.