I have a report that I send out to clients reporting their back links as displayed by Yahoo's link: operator. (yes, I do note to the client how this is just directional, and that this is not an entirely accurate figure).

I filter to backlinks from any other domain, to entire site.

However, I'm curious how you would do it. I usually do link:www.example.com instead of link:example.com. The numbers are always different.

Why not simply use Yahoo Site Explorer (siteexplorer.search.yahoo.com)?

doing the link operator takes me to Site Explorer.

Oh, now I see what you're asking. Basically, www.example.com is considered a subdomain of example.com. Apparently, for the sites in question, some backlinks use the www version and some the non-www version, accounting for the different numbers.

Right. So Should I be sharing the links for the www, the non-www, or both of them in aggregate?

If they were 301 redirecting the non to the www, then maybe I could combine them and feel good about that number.

I could run arguments every which way in my head on this one and not come up with one that can't somehow be refuted.

I don't think I'd simply combine the numbers for the www and non-www versions because I think we can assume there would be at least some dupes in there, leading to inflated numbers.

I might use whichever version the site uses, i.e., if the site is canonicalized to redirect all non-www requests to www requests, I'd go with the www backlink count. This would still not be an accurate count but, as you stated above, be a "directional" data point.

Now, what I *might* actually do if I had the time (or inclination), is to try to calculate a baseline percentage of duplication between the two versions, then use that "fudge factor" going forward. For example, you find that 10% of backlinks using the non-www version are dupes. For each report period you would simply add...

www version + (non-www * 90%)

...or some such. Then maybe re-run your baseline analysis say every 6 months or so.

But I'm not sure if that would be worth the effort or result in any more meaningful numbers.

>If they were 301 redirecting the non to the www, then maybe I could combine them and feel good about that number

If this is your client, the first thing you should do is get the 301 in place. Then you could feel good.

It sounds like "link:example.com" is really an alias for "*.example.com" if that makes more sense.

Whereas "link:www.example.com" and "link:blog.example.com" restrict it to the sub domain.

Like jimbeetle said, people may not always use www.example.com to link to the site.

Its my personal experience that www.example is better for this reason, then trying to 301 everything to just example.com

Exactly 301 requires, this will avoid content duplication issue.
As google treats both www or with out www URL' as different URL's.
Why to take chance. And there after no matters if you build links with www or with out www.
Hence i would suggest with www.