Just ran a quick report today on SEOmoz to see how many unique domains sent the site traffic vs. how many unique links Yahoo! reports the site has. Interesting numbers:
- Yahoo! reports 12,600 links, but if I scroll to the end, that number drops to 7,040.
_ - Google shows 89,700 mentions of SEOmoz (there were 0 two years ago), of all the pages I checked (which was far fewer than 89K, all were linking to the site). I scrolled to the end with them and came up with 451 results, before GG asked me to "repeat search with omitted results included". At that point it showed 89,400 results.
_ - My analytics data shows that since Sept. 1, we've had 42,311 unique domains send traffic to the site
_ - As for unique referring linking pages, there's 46,610.
So... Yahoo!'s linkdomain is pretty far off, as is Google's number of mentions. I'd like to try to integrate link tracking with visitor analytics, just to see how well the search engines pick up new links. The current link data is really only of value in a relative sense (i.e. comparing one site against another).
I can tell that the figures are pretty similar to what I see / know for our sites. If you check some of the missing pages you'll most likely notice SE's haven't indexed them; so this is due to indexing / visibility issue on their part. Especially this hits for natural backlinks like personal homepages, bookmark pages etc.
I wonder how accurate the page impressions are that are harvested by Google? My logs and Google do not always seem to jive.
I also question the undeniable gap between Yahoo’s count, Google’s count, and MSN’s count for their related search? I guess this is why so many believe statistics can be subjective, or that you must prove results from so many sources in order to incur credibility.