Robert_Charlton - 10:02 am on Jan 26, 2005 (gmt 0)

Over the past several years, there have been numerous forum discussions about the problems that excessive crosslinking can cause.

Some of the discussion has come from the perspective of using crosslinks to give other sites a PR or anchor text boost from a common network, with most questions about "how much crosslinking can you get away with?"

I have a question from another perspective. For some sites of company subsidiaries, I want to keep the crosslinking for users and... to avoid crosslinking complications... get rid of it for PR and anchor text. I'm thinking about the new "rel nofollow" attribute to partially unrelate the sites.

Some background for those who may not have observed it... as Google has come down hard on crosslinked networks, there's been a fair amount of collateral damage, in my opinion, to sites under one ownership that crosslink for business and navigation reasons, rather than for SEO reasons.. These generally are sites for different brands or product lines or locations, but having some commonality.

What I've seen is that Google often appears to be "clustering" the results of some sites that it sees as related or similar, dropping out returns from all but one of the sites on search queries that in any way overlap.

Different imprints of a textbook publishing company might be a good example. One imprint might be for science textbooks, another for law textbooks, another for history. They're all established as separate companies. They have different public images. But the separate sites for these may all be hosted on the parent company's server, and because they all share the term "textbooks," all but one might disappear.

The site commonalities may often go beyond simply crosslinking. There may be hosting or IPs in common, nameservers in common, similar patterns of inbound and outbound links, or related search targets.

Assuming some of these other issues are taken care of, and the sites exist more or less independently with independent inbound links, I'm wondering whether the new "rel nofollow" attribute might help in maintaining navigation between the sites while preventing at least some of the clustering difficulties.

Or, might this be an inappropriate use of the attribute that could in fact backfire?

Or, might the attribute not be necessary if the sites do have sufficient independent inbounds.

Lots more to be said about the above, about why companies have different product lines and locations in the first place, etc, and how this may also relate to Google's desire to cluster... but given that this is the way many companies are built, can "rel nofollow" help here?

Related reading...

Similar sites on same IP address
Ranking issues
[webmasterworld.com...]