Like many others, I occasionally use Compete and Alexa data to compare traffic of websites whose logs I don’t have access to.  I know these services are imperfect, but a comparison I ran today of NYTimes.com vs. USAToday.com vs. WashingtonPost.com shows just how anecdotal the data from these services is.

Below are the results of a comparison of the three sites from Compete for the last year.  According to Compete, USA Today gets more unique visitors than the New York Times.  My gut tells me this is completely wrong.  While I’ve read my share of issues of USA Today on airplanes and at hotels, it is unfathomable to me that its website is more popular than NYTimes.com.

So I went over to Alexa and did a similar comparison to dramatically different results.  The New York Times is well ahead of both the Post and USA Today in terms of Daily Reach, with USA Today actually trailing the Post by a small margin. To put this in perspective, Compete is reporting that USA Today gets around twice the traffic as Alexa is reporting.  So this discrepancy isn’t within any normally accepted margin of error.

To me, the Alexa result feels rights, and a look at the Nielsen data makes me semi confident that my gut instinct is more correct than Compete.  However, the Nielsen data routinely comes under criticism itself, as does that of competing service Comscore.

The fact of the matter is that the only way to track site usage with any precision is through log files, which the firms providing these numbers don’t have access to.  No one has truly figured out to track overall web usage accurately as of yet, so we should all remember to take these numbers with a huge grain of salt.  Even when the the stats you get back match your working hypothesis, as the Alexa numbers do in my USA Today vs. NYTimes vs. WashingtonPost.com comparison.