Interesting recent article on the murky business of measuring web traffic from Carl Bialik of the Wall Street Journal. He elaborates on some points in a post on The Numbers Guy blog. Most interesting to me is his discussion of the gap between internal site numbers from web analytics tools (e.g. Google Analytics, Omniture) and competitive analysis tools (ComScore, Nielsen, Compete, Quantcast), with the latter’s numbers much lower than the former’s. My two thoughts:

1) While bothersome, this gap is much less of a problem if everything is “directionally consistent”. If Compete shows that my site gets 60% as much traffic as Google Analytics does, I want this ratio to be true for all of my competitors. If Google Analytics shows my site’s traffic going up 50% over the last year, I want Compete to as well. I think part of the frustration among website managers is that, not only are there big gaps between internal and external counting, there are also inconsistencies in these other areas.

2) If cookies create problems with double-counting users who visit sites on multiple computers, can’t we focus more on visits and less on unique visitors?