[WEB4LIB] re: public library usage statistics

[Alnisa Allgood]

My day for email responses...

I highly recommend FunnelWeb, which I believe is free now for the basic version. For a visual of a report see http://nahic.ucsf.edu/stats/2004/11_November/  

One of the things it allows you to do is configure items for exclusion. For example for this client, who wanted to track the response for a recent push for a new document for the NIIAH, I set exclusion rules that consisted of making sure no one inside their organization was counted towards the final statistics (they have a 100 of the 250+ IPs associated with a subnet, so I just excluded the subnet, I excluded any links to the CSS, images, or other administrative directories, excluded my IP, excluded robots, etc. and then added a few inclusions, such as monitoring the downloads director, etc.

They are primarily interested in downloads of files related to a single project, but did also want to see other web site stats, with a particular focus on what "types" of organizations (i.e. how many other universities vs nonprofits were using their data), but I found the control levels in FunnelWeb to be excellent. I feel like I haven't even scratched 5% of the programs capabilities but the stats I produced for them was soo much cleaner and clearer then their WebTrends report or the report from a few other tested analyzers.

Obviously the list of things you can't really know is long and complicated, but at least for the things you can know or things that you you can take a brief snapshot of can be done well.

Alnisa

-----Original Message-----
From: Robert Sullivan <rsullivan at sals.edu>
Sent: Fri, 04 February 2005 13:35:17
To: Multiple recipients of list <web4lib at webjunction.org>
Subject: [WEB4LIB] re: public library usage statistics

> Before you decide what stats to gather and how to present them, take a
look
> at the latest edition of Stephen Turner's "How the Web Works," at
> http://www.analog.cx/docs/webworks.html as well as Jeff Goldberg's
"Why web
> usages statistics (are worse than) meaningless," at
> http://www.goldmark.org/netrants/webstats/

Those were interesting.  We were more interested in knowing what
visitors actually used rather than an inflated figure to impress the
authorities, so I brought the logs into a database (I used Visual
FoxPro, but you could probably use Access or another programmable
database) and removed the entries for CSS files, graphics, robots.txt
etc.  I also remove entries from spiders automatically.  After some
manual browsing for the less-obvious mechanical visitors (a lot of hits
from one address in a very short time), I come up with a reasonable
estimate of "pages viewed by a human" which is still subject to the
vagaries of caching but gives us a sense of whether we are devoting our
resources in the proper areas.

Bob Sullivan  
Schenectady County Public Library (NY)  
Schenectady Digital History Archive