We use two products because each has its strengths.
 
a.  Ancient (4.x) version of WebTrends.  
Pluses: 
-  Since we have been using it since the last millennium, it's the
closest we can get to accurate trending over time on the usage of our
server.  
-  Unlike later versions of the same product, the license doesn't max
out after processing X number of page views.  So at this point it's
free.
-  Counts everything in the log files, so if I edit out certain lines
in the log files and rerun the reports, I can know % of usage from
in-house, Googlebot, Slurp, etc.
Minuses: 
-  Can no longer report on things like referring search engines (since
it's so old it doesn't know Google is a search engine).
-  Reporting capabilities are not as slick as Google Analytics.
-  Counts everything in the log files, so it's got a lot of noise data
if you just want to see what real non-bot users from outside the library
are doing.
 
b.  Google Analytics
Pluses:
-  Very nice reporting.
-  Easy setup, including configuring it to exclude internal use.
-  Free.
-  Requires JavaScript, which more or less automatically excludes
non-human bot hits from the reports.
Minuses: 
-  Someone else has our log data.
-  Haven't found any feature to combine multiple logs and report on all
3 public servers in a single report.
-  Requires JavaScript, which excludes reporting on usage from browsers
with JavaScript turned off/unsupported.
 
I looked around a number of other tools and found many of the exact
same pluses and minuses, as well as the rather staggering price of some
of the commercial offerings, many of which now also use the
JavaScript-ping-to-remote-server method.  Currently, I actually have
somewhat more faith in Google to keep our users' data private than I do
in the pricey analytics vendors, but this may be naive on my part.  At
any rate, using one log-based tool and one JavaScript-based tool is
working OK for us for now.
 
There's an archived webcast about analytics options specifically for
measuring web "visits" from outside the library here:
http://www.infopeople.org/training/webcasts/webcast_data/255/index.html

 
 
Genny Engel
Sonoma County Library
[log in to unmask] 
707 545-0831 x581
www.sonomalibrary.org 
 


>>> [log in to unmask] 11/11/08 12:52PM >>>
Library Code People:

1 - What do you use for your web statistics package?  Are you happy  
with it? Pros/Cons?

2 - What do you wish you used or had access to?

3 - Opinions on Specific Projects:

3.1 Piwiki/Mint

Piwik and Mint both seem pretty interesting to me because they solve  
some of the problems of "traditional" log file analysis (see
http://www.boxesandarrows.com/view/the-limitations-of#comments) 
, while of course introducing their own set of problems:  given their 

reliance on a RDBMS to store each page load, there are some obvious  
scaling concerns for very high traffic sites, for example.

I wonder if anyone here has put either of these or similar systems to 

the test on high traffic (define in your own terms) sites.

3.2 Google Analytics and/or Urchin

Some libraries have incorporated Google Analytics into their privacy  
policies:
http://www.google.com/search?q=google+analytics+libraries+privacy&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a

  .

So, anyone here passionate one way or the other?  Other Pros/Cons?

Of course, favorite resources, questions I should be asking and the  
like are welcomed and appreciated as well :).

In advance:  thanks!

Cheers,
-Chad