Dave Winer aside, large numbers of bloggers can charitably by described as obsessed with web statistics. People make all sorts of claims based on their reading of them. When they’re up, its “Top of the World, Ma!” When they’re down, it’s Hunter S. Thompson time.
It’s a slim reed that bears responsibility for such behavior, especially given the fuzziness inherent to creation of such statistics. Let’s use the stats for fellow UNC blogger Eric Muller’s IsThatLegal as an example. He uses Sitemeter to track visitors and page views, mostly because it’s free. We do the same here at Hraka. Here’s his sitemeter page. Right now it’s showing an average of 1,198 visitors a day.
However, since Isthatlegal.org is hosted on our UNC web servers, we count visits to it as well, based on the logs the web servers generate each day, using a web stats program know as Urchin. There’s quite a difference, as Urchin claims that 1600 more people comprise Eric’s daily readership than does Sitemeter.
As Eric asked me when I pointed this out to him, why does Urchin register so many more visitors than Sitemeter?
Short answer: Because Sitemeter is free, and Urchin is not, so presumably Urchin is therefore better. I’m not the only person to think this.
Sitemeter’s Answer: It can sometimes be difficult to compare two different tracking systems. First, you need to make sure that you are comparing apples-to-apples. If you are looking at the number of page views on Site Meter, make sure you compare it to page views on the other tracker. If you are comparing visitors between the two systems, make sure the definition of a visitor is the same on both systems. Some trackers will count a person who visits a site multiple times in the same day as a single visit.
Longer answer: The Sitemeter answer is kind of a copout, as “visitors” and “page views” are fairly standard terms. Differences in results between web stats programs are more likely to be due to differences in the algorithms each uses to track visitors. In any case, our Urchin setup tracks both actual visitors–though it uses the term “sessions,” and page views.
It’s important to realize that all web stats counters count differently, and all of the numbers one gets from them are estimates. Much depends on how they are set up to count visitors from AOL and other large ISPs, where a reader may or may not have the same IP address throughout a session. That’s important, as the behavior of an IP address over time determines the number of visitors to a site.
Sitemeter and Urchin use different methods to count visitors, so they come up with different numbers. Sitemeter is code based. One has to add a snippet of code to the page one wants tracked. That code updates a counter whenever a page is loaded. Sitemeter also uses cookies to try and prevent itself from counting the same person twice.
What else those cookies are used for, God only knows, so many people end up blocking them, which means that Sitemeter will always either miscount or undercount that type of visitor. What the use of cookies also means is that readers who do accept them are only counted once a day, so someone who visits at 10 in the morning and then again at 10 in the evening is only counted once.
Instead of relying on a code snippet and a continuously updating counter, Urchin’s stats are created only after the previous day’s server logs have been parsed–calculating up all the UNC webstats takes up quite a large chunk of the early morning hours, in fact.
How Urchin parses the logs is important, for it is critical to the difference between Urchin and Sitemeter. Urchin uses the term “sessions” instead of “visitors.”
From Urchin’s point of view, the number of people visiting a site is not nearly as important as the number of times people visit the site. Here’s how a session is defined;
A Session is a series of hits from one visitor (as defined by the visitor’s IP address) wherein no two hits are separated by more than 30 minutes. If there is a gap of 30 minutes or more from this visitor, an additional Session is counted.
With Urchin, the theoretical reader who visits at 10am and 10pm is counted both times, though he is not if he visits at 10 am and then again 10:19 am. Given that blogs are meant to be read multiple times a day by multiple readers, Urchin’s idea of a “session,” is probably more useful than that of a visitor.
Consider, which would be more valuable to potential advertisers if Eric were to join BlogAds, the one visitor/one visit that Sitemeter records, or the one visitor/two visits that Sitemeter records?
The problem for most bloggers is that Urchin must be paid for, so the fact that it counts more efficiently is kind of a moot point. Those whose domains are with Hosting Matters have access to Webalizer for their sites. Presumably those with other ISPs do as well. It appears to count in a similar manner to that of Urchin. I’d provide a link to ours, where the number of “visits” calculated by Webalizer greatly exceeds the number of visitors as shown by Sitemeter, but there doesn’t appear to be a way to do that at the moment.
Which is a problem, because if we were to ever join Blogads I would much prefer to charge for ads according to those numbers, but short of handing out my password, there’s no way to justify the higher rate–though in the great scheme of things, that rate would still be virtually nothing.
