Patrick Ruffini has discovered a SiteMeter flaw that over-estimates the number of unique visitors to a site.
His hypothesis seems to be corroborated by looking at other sites, especially Michelle Malkin’s site (it’s lucky for this test that she redesigned her site in a unique manner to support Ruffini’s claims).
Patrick noted that SiteMeter only counts up to 100 visitors (for the basic, free service– the premium stores up to 4,000). After that, if visitor 101 clicks on a link on Kos, it says it’s a unique visitor. On a site like Kos, it’s entirely possible to have visitor 101 click on a link on Kos after 1 minute and be counted a unique visitor.
This flaw is just not for Kos, but for any high-traffic site using basic SiteMeter. If you look at Michelle Malkin’s Sitemeter track for the past 12 months, you’ll see a jump in page visitors and page views in June, when she instituted her new format. Malkin gets a lot of traffic (sadly, not as much as Kos), but she gets enough to verify the hypothesis that the site-meter clicky problem exists. Note that to access a story on MichelleMalkin.com that’s not the headline story, you need to click on the sidebar link. This generates more click-throughs, and as a result, a greater number of unique visitors. If you do the math, she gets approximately 1 visitor every 1.6 seconds, so her “flip time” is roughly 62 seconds. So if a reader browses the first story for just over a minute, then clicks on a side story, that reader is a new visitor, according to SiteMeter.
Now, if you look at Kos, he gets 450,000 visitors a day, which is about 5 visitors a second. That means every 20 seconds, someone clicking at Kos’ site is a new visitor. In this case, the bigger your site, the bigger the statistical inflation of your # of visitors.
The SiteMeter FAQ seems to dispute this, but the evidence gathered here seems to be contrary to that belief. An additional test to verify this hypothesis would be to make the two SiteMeter pay sites (which allow 4000 stores referrals) and see if Kos and Malkin experience a decrease in visitors but a retention in page views.