[Web4lib] web page statistics

John Fereira jaf30 at cornell.edu
Wed Dec 21 07:32:39 EST 2005


At 08:21 PM 12/20/2005, Bret Parker wrote:
>And we have had success with Analog.
>  http://www.analog.cx   It is fast and once you get used to editing 
> the config file, there are some great things you can do with it. It 
> is also free.

I also liked a tool simply called "visitors" 
(http://www.hping.org/visitors) as it was really easy to set up and 
could produce a variety of custom logs from the command line.

It is worth noting that all of these tools (visitors, 
analog,  webalizer, etc) can only get information that is captured in 
the web server access logs.  For many of the web applications I've 
developed which use a database backend I capture additionally 
information using the application.  For example,  one of the systems 
I developed has a patron registration/login page.  It updates a 
timestamp everytime a new patron logs in or registers in the 
database.   From that information, using a simple SQL query I can 
determine how many new patrons have registered in the past month (and 
every month since the system was released), what percentage of users 
register but don't continue to use the system, and other metrics that 
are of interest that just can't be captured in a web access log.  The 
tricky part is predicting what statistics will be required so that 
they can be captured programmatically from the beginning.  I don't 
know how many times I've received a request for some metric about the 
system a year after it was released and been unable to provide the 
statistics because they data hadn't been captured.



John Fereira
jaf30 at cornell.edu
Ithaca, NY 



More information about the Web4lib mailing list