More than a few sites have commented on Michelle Malkin’s post on how many people are reading political blogs. I decided to compare the to services that have been mentioned (Alexa and Site Meter) with one that has been overlooked – server logs. Most hosting accounts come with statistics reporting packages such Awestats or Webalizer which give highly accurate traffic statistics because they use the web servers own log files to generate their reports.
I ran the numbers for these three measures for February 2005 at Wizbang to see the differences in the three methods of measurement [Stats shown after the break]. February was a bit of a down month compared to January 2005, but I’m missing a 10 days of server stats for January. I extrapolated the daily average to those days 10 days to estimate the total visits for January, then generated the same stats as shown below for February. The differences in traffic measurement between the three methods was (as a percentage) almost exactly the same as for February.
A couple notes:
Alexa rankings are estimates pulled from the universe of Alexa Toolbar users then extrapolated to the internet in general. Alexa numbers are particularly useful for comparing the traffic of two sites to each other and in generating traffic rankings because all sites are measured in a uniform fashion. Using Alexa reach per million estimates to try to generate statistics for daily page views is basically a SWAG**, as both the reach and the number of daily internet user are estimates.
Here is what Site Meter tracks: Site Meter tracks page views and visits. You may also have heard the term “hits”. When someone comes to your site, they generate a “hit” for every piece of content that is sent to their computer. Viewing a single web site page would generate one hit for the page and one hit for every individual graphics file that was on the page. A single page could easily generate a dozen or more hits. When you are browsing a site, every time you follow a link, it is treated as a single “page view”. Site Meter defines a “visit” as a series of page views by one person with no more than 30 minutes in between page views.
Awestats and Webalizer numbers will almost always be higher than Site Meter numbers since it does not discard visits from the same IP address less than 30 minutes apart, as does Site Meter. In this regard Site Meter’s methodology is probably more accurate than Awestats and Webalizer. Additionally the host based reporting packages capture RSS/Atom page loads where Site Meter and Alexa cannot and do not. In a typical month at Wizbang the RSS/Atom feeds are anywhere from 15% to 25% of the server measured page views. As all but one of the feeds includes full content, the majority of those page loads are never captured as visits by Site Meter. After adjusting for the feed issue (~5,000 visits a day) the server generated statistics and the Site Meter statistics are much closer.
Awestats – February 2005
Unique visitors – 344,302
Number of visits – 659,646 (1.91 visits/visitor)
Page view – 1,653,716 (2.5 pages/visit)
Hits – 11,346,005 (17.2 hits/visit)
Bandwidth – 126.60 GB (201.24 KB/visit)
Average visits per day 23,559
Site Meter February 2005
Site Meter visits – ~500,000
Site Meter page views ~ 640,000
Average visits per day 17,857
Alexa February 2005
Alexa Reach Rank – 90 per million
Average visits per day 10,800