use log analysis

[Roy Tennant]

I can't think of a time when I deleted a page based on low use, but I can
give some examples of how I used what I've learned by monitoring my logs
to change what I'm doing.

I have a program based on Thomas Dowling's browser statistics program that
takes the last 10,000 hits on my server and divides them up according to
what section of the server is being accessed or, in the case of errors,
into the type of error. You can see this in action at
http://sunsite.berkeley.edu/cgi-bin/snapshot.pl . This allows me (not you,
since it requires authorization) to see where people are failing (that is,
the "404 not found" errors). Often this is because they have used a
variant of the path to a file. In UNIX, it is trivial to put in an alias
so that various versions of a path will work. for example, all of the
various URLs now work on my server:

sunsite.berkeley.edu/InternetIndex/
sunsite.berkeley.edu/internetindex/
sunsite.berkeley.edu/Internetindex/
sunsite.berkeley.edu/internetIndex/

Why, after all, should the user fail when they are so close?

A while back I also started monitoring the search strings entered on our
server-wide searching by simply writing them out to a file. The first time
I looked at it I was horrified to discover that most people didn't have a
clue what they might find. So I rushed to create a file describing the
coverage of our server-wide index, and a description of those databases we
have that must be individually searched. This is available at
http://sunsite.berkeley.edu/Help/coverage.html if you want to see it.

And of course, thanks to Thomas Dowling, I also monitor what browsers my
users are using at http://sunsite.berkeley.edu/cgi-bin/browsercount.pl .
This I do to understand the impact of any new technology I may want to
use.

I am a strong believer in monitoring what is going on with your Web server
and how your users are using (or not using) it. What I've found is that
often it *will* lead to changes if you pay attention.
Roy

On Wed, 11 Feb 1998, Cal Boyer wrote:

> The recent article in "College & Research Libraries" on 'recommendations
> for benchmarking web site usage among academic libraries' prompts the
> following request:
> 
> For those who have used any form of log analysis of your web server
> activity as a basis for making changes in your web pages or web
> offerings, would you please note or describe briefly on WEB4LIB some
> examples of your changes/improvements.
> 
> Thanks in advance.
> 
> Cal
> 
> Calvin Boyer
> Longwood College
>