[WEB4LIB] RE: Reporting Web usage: what numbers do you use?

Thomas Bennett bennetttm at appstate.edu
Tue Oct 26 12:12:04 EDT 2004


  All of our PCs have a unique ip address.  Though, all PCs in the
library end in library.appstate.edu which would be easy to pull stats
from.  In fact all public PCs begin with set of letters and a unique
number like abc123.library.appstate.edu, abc246.library.appstate.edu,
etc.  So public PC data could be pulled from the log file for specific
data.  The rest of campus is simliar for instance xxx.cob.appstate.edu
(College of Business) where xxx is a name or number.

  There is a catch with your awstats and unique vistor but probably not
enough to really skew the results.  Say I have a dialup connection and I
connect with a.b.c.d and disconnect in 30 minutes.  Someone else
connects after I disconnect and gets a.b.c.d.  If both of us went to
your WEB site then the second vistor would not appear unique if I read
your email correctly.


On Mon, 2004-10-25 at 19:59, Lori Ayre wrote:
> More questions pertaining to Karen's dilemma....if a site doesn't want to
> place cookies on their visitor's machine and are relying on the visitor's IP
> address, how can they reliably track the number of 'unique visitors' to
> their website? 
> For example, awstats increments the 'unique visitor' count each time an IP
> address hits the site but it doesn't increment that counter again until the
> minimum VISIT_TIMEOUT is reached which, by default, is set to 10000 seconds
> (they call that hour).  
> Seems to me that this number isn't very useful because I'm assuming everyone
> from the same library is going to hit a site with the same IP address (their
> public IP address).  My assumption is that most libraries use a firewall
> that performs a NAT process and displays one IP address to the world. 
> I'm wondering if my assumption is correct.  *Do* most libraries only use one
> public IP address such that whether a staff person or 20 people on your
> various public access computers hit a site, it will all look like traffic
> from the same ip address?  
> And if yes, what's a better way to figure out how many of 'your people' are
> visiting a site?
> Lori Ayre
> -----Original Message-----
> From: web4lib at webjunction.org [mailto:web4lib at webjunction.org]
> On Behalf Of John Creech
> Sent: Monday, October 25, 2004 3:19 PM
> To: Multiple recipients of list
> Subject: [WEB4LIB] RE: Reporting Web usage: what numbers do you use?
> On Mon, 25 Oct 2004, K.G. Schneider wrote:
> > The other is to generate some discussion about *why* you use what you 
> > use. In particular, I have found that statistical packages often have 
> > wildly variant rules about what a "user session" is.
> One thing folks may want to consider, Karen, (and I think sometimes they 
> forget this or don't know it) is that some of these packages are quite 
> configurable and you can get a variety of details and snapshots of what 
> web servers collect.
> John Creech, Systems Librarian
> Brooks Library, Central Washington University

More information about the Web4lib mailing list