Canonical links across domains - Good, bad, indifferent?

[Marshall]

When you have a site with multiple domain names, does it hurt to use a canonical link across them? I see in Google's "Official" Webmaster Blog [googlewebmastercentral.blogspot.com] that they will honor it

Can this link tag be used to suggest a canonical URL on a completely different domain?
**Update on 12/17/2009: The answer is yes! We now support a cross-domain rel="canonical" link element.**

but I am curious if this is a good or bad SEO? In this context, the question applies to an e-commerce site.

Thanks in advance.

Marshall

[tedster]

I think cross-domain canonical links are often a bad idea. First, only Google supports them. And in most cases I've seen, there is a better solution that doesn't depend on Google getting it right.

That said, I suppose it can direct Google to which version of cross-published content should be indexed.

[Marshall]

First, only Google supports them.

That was a concern wondering how other search engines would treat it.

there is a better solution that doesn't depend on Google getting it right.

Such as?

Here is the issue with one site I manage. There are multiple domains pointing to it and they have specific landing pages. However, they share the same dynamic shopping cart though they all cover different product areas. The shopping cart itself is dedicated to one of the domain names to be https verified. Problem is, WMT shows one domain having 5,000+ links to the other, but in essence, they are all the same domain.

I guess I could use a UA detector and document.write when Google visits, but is that really practical.

Marshall

[tedster]

I'm not 100% clear on the set-up you describe. Do different domains use the same shopping cart system which then can serves visitor the same content (products) on different domain names?

Also, would this be a preventative measure for your sites, or are you already seeing real ranking and/or traffic problems in practice?

[Marshall]

Hey tedster, here's the set up.

My client has an e-commerce site that for years had only one domain name, widgets.com, and ranked really well.

Last spring, she decided to add brand specific domain names to the site, we'll say BrandAWidegt.com, BrandBWidget.com and BrandCWidget.com. These names have specific landing pages different than widget.com's home page. The products across all domains are the same, so if you look at a product detail page for any product, it will start with the domain name the visitor came in on. To this point, all is well and good.

Not long after this, widgets.com fell way down the SERPS from page 1 top five to page 10 or better. The new domain names generally rank on page one. That being said, I'm not sure if the problem here is the different domain names or Panda update.

When a customer goes to check out, the cart defaults to widgets.com to establish a secure connection as it is the only name that can be used in the cart's script. However, if the person continues to shop, the domain name will stay on widgets.com which is why I think WMT says one of them has 5,000+ links back.

After writing all this, I'm starting to think the problem is in the two different navigation settings which could give the impression of duplicate content. If that be the case, I can write a server side detection script based on the HTTP_HOST to generate navigation appropriate to the specific domain name.

I guess my biggest concern is not so much the fall of widgets.com in the SERPS since the other domain names make up for it. It is whether or not this can be seen as duplicate content or even some black hat SEO, which it is not. This is a similar concern to the one someone raised in another thread about putting mobile specific pages on a site which, while designed to be mobile friendly, may be seen as duplicate content.

Probably the easiest thing is make a base url and say the heck with it and see what happens. But I really do not like that idea.

Am I making any sense? :)

Marshall

[tedster]

Well, the first thing is to be balanced in understanding what "duplicate content" means, and which kind of duplicate content is actually a problem.

There is all manner of duplicate content on the web and a lot of it is innocent or even accidental. In most of those cases, Google will simply choose which version to rank and they filter out the rest. No penalties or loss of trust occurs to the domain because of this multiple use of content. This sounds like what happened to the family of domains you described.

Sometimes a domain does get into trouble with Panda if it is seen as mostly duplicating content from other domains. So far, Panda troubles get doled out around certain specific dates reference dates [webmasterworld.com] so that makes it pretty easy to spot Panda problems compared to other ranking issues.

If that be the case, I can write a server side detection script based on the HTTP_HOST to generate navigation appropriate to the specific domain name.

From my abstract distance, that sounds like a promising direction.

[Marshall]

tedster,

Thanks for the input and letting me bounce ideas off you. I think the server-side script (asp response write) to generate the appropriate navigation menu is the way to go. I did notice that the shopping cart software generates some absolute url's to the primary domain which may also be part of the problem. Guess I'll do a little rewrite there too.

Oh how I long for the days when things were simple.

Marshall