I read about a website called Compete. It provides comparative analysis of website traffic. After trying the obvious Google vs. Yahoo vs. MSN and seeing Google and Yahoo as very close competitors (which was a little bit surprising), i thought about trying something less obvious.
Usually, websites in foreign languages are not tracked by this kind of websites, but I decided to try. My comparison was the three major Israeli newspapers: Ynet vs. NRG (Maariv’s website) vs. Haaretz. I know that at least Ynet and Haaretz have separate URLs for HE and EN versions, so what i compared were the HE URLs. Also, Ynet is considered to be by far the leading online outlet in Israel, though i am not sure who holds the second place (in print that would Maariv). Now, imagine my surprise when i got this:

According to this graph, Haaretz is by far the most visited (in terms of unique visitors) online outlet, Ynet is only second, and NRG (not surprisingly) the third. But that is not all. Here is the ranking comparison:

Again, Haaretz is in the lead, followed by Ynet (closer this time), followed by NRG. Though i am not sure how to read this metric. On the one hand it ranks sites based on the unique visitors metric, but the frame of reference is “the top one million websites in the U.S.”
Where it is getting more intuitive is in the comparison of the number of visits:

But even in that case, the gap between Ynet and Haaretz is not that significant, yet it is growing. Same goes with the length of stay (in minutes):

It looks like people tend to spend more time on Ynet and NRG, but not on Haaretz. That is, again, counter-intuitive, because Haaretz usually has longer and more complex articles as opposed to sound bites on Ynet and NRG. One explanation to that metric can be that people arrive to Haaretz’s homepage, but do not actually read. Or, alternatively, there may be more content on the homepages of Ynet and NRG, compared to that of Haaretz. So, people navigate away from the homepage to read the articles quicker on Haaretz.
In fact that is one of the more confusing parts for me in these metrics. Do they account only for activity on the exact domain or on all its sub-domains as well? Also, it is quite possible that many people have domains of their favorite newspaper saved as a bookmark or they choose it from the auto-complete line in the address bar. In that case, some of the confusion with Ynet can be explained, as its homepage URL in auto-complete looks something like that: http://www.ynet.co.il/home/0,7340,L-8,00.html instead of a simple www.ynet.co.il (even though i just saw that this is not really the case with the new Firefox).
However we are not done with surprises yet. Here is a comparison of monthly attention each on the of website was getting in the last year:

This one is, again, a relative metric of time spent on a given domain with the refernce frame of “the total time spent online by all U.S. internet users,” which i find quite a confusing one.
Number of pages per visit is yet another surprise:

It seems that people tend to browse through NRG more than through Ynet. However, this metric implies that they track not just the homepage, but also the rest of the inner pages, which highlights the puzzling aspect of previous results.
All in all, i found this excersise interesting, yet very confusing as it contrudicted the common wisdom i held so far. This is why i kept on reading in order to find out how exactly they are getting their data. So, i think my confusion was partly resolved when i read that they derive their information from a “sample of 2,000,000+ U.S. internet users” who gave them “permission to analyze the web pages they visit and ask them questions via surveys.” So, on the other hand, that may explain the difference in popularity of the various outlets, but on the other hand, i am surprised that the Israeli websites in HE actually made it into their analysis (ranked among the first 60K out of a million websites). So, i am still puzzled.