Cross-Domain Canonical - Should I use it under the following circumstances?

[cmaddison]

I have a number of hyper local directories, where businesses get a page dedicated to them. They can add images and text, plus contact info, etc.

Some businesses list on more than one of these directory sites, but use exactly the same description.

I've tried asking businesses to use unique text when listing on more than one site to avoid duplication issues, but this is proving to be too much work for the business owner!

Can I use a cross-domain canonical and point Google towards the strongest domain from the group of directories?

What effects will this have? And is there an alternative way to deal with the duplicate content?

Thanks - I look forward to hearing your ideas!

[Dr-Pete]

It's always hard to talk in generalities about complex issues like this, but it sounds like a situation where cross-domain canonicals might make sense. I guess it really boils down to whether you're having issues with the duplicates and what the scope is (are there 3 of each or 300). In some cases, those duplicates just mean that one site will win, and Google will pick the winner. In other cases, the main site could actually be harmed by the duplicates. In some cases, honestly, multiple sites might rank fine. It really varies wildly.

The cross-domain canonicals would help prevent any kind of duplicate penalty (like being hit by Panda), but it would also mean that the non-canonical versions would no longer rank. So, you'd be protecting the strongest site for each listing, but possibly cutting off the smaller sites.

I haven't seen an implementation where different sites were canonical for different listings/articles/etc., at least not on large-scale, so that's a bit tougher to predict. If you have sites A-Z, and A is canonical for one listing, B for another, C for another, etc., that could get a bit tricky. I know large organizations, like newspapers, who syndicate content, have had good results in many cases with cross-domain canonical.

There is also a syndication-source tag, but that's really a weaker tag, and I haven't seen much data on it. The other option, traditionally, would be a solid link-back strategy (the non-canonical versions link to the canonical version). Unfortunately, at large scale, that could start to make you look like a link network, so I think that gets risky in this case.