Summary - Reporting web statistics

Thu Jul 1 08:19:30 EDT 1999

Hi!  I wanted to thank those who took the time to answer my query on
reporting web statistics.  Someone asked for a summary of the answers I
received.  Here they are!  If you have anything to add, please do!

Lynn
--
B. Lynn Eades
Distributed Learning Librarian/WebManager
Health Sciences Library
CB# 7585
University of North Carolina at Chapel Hill
Chapel Hill, NC  27599-7585

Phone: (919) 966-8012
Email: beades at med.unc.edu

-------------- next part --------------
We have just begun using WebTrends.  (In fact, we have tested it on six
months from last year.  We will begin use with this fiscal year, July 1.

Thanks!

Grace
____________________________________
We use WebTrends for our stats.  You can see examples at:
http://library.boisestate.edu/statslink/  or
http://www.lili.org/stats/

In both cases we only show a small subset of the stats that are available, 
and in only a few of the ways it can be presented.  I don't work for 
WebTrends, but highly recommend it.  The price is pretty reasonable if you 
remember to ask for an educational price.  You can download a trial copy 
from www.webtrends.com to see if it will do what you want.

As to the question of how management uses the data, I'd say they don't 
really.  I provide it, as I have done for almost two years, on the 
assumption that someday they'll care and need it urgently.  It'll be there 
waiting.

Dan
___________________________________________
Here's our praxis:

Once a month all the "active" users of our server (those
entitled to publish on it) receive a report on overall
hits.

In addition to this anyone publishing on the server receives
a report concerning "his" documents as often and as detailed
as he or she desires it.

The general monthly report each month contains 2 to 5 extra
sections investigating into and documenting various more
special aspects (development in the user-base and their 
interests, surfing behaviour, changes in load for special
sections, most popular documents, who-what analysis, etc.,
etc.) Giving them all the detail every month most probably
would mean information overload for them, but like this
they come to learn about what can be learned from such 
statistics, and can formulate on wishes for "special
statistics".

BTW: There seem to things that are apt to increase publishing
participation: a) information about the hits for the documents
of the single author and b) information about hits from countries
whereto the author has not been yet for vacation ... [;-)].

Could you summarize the results of your enquiry for web4lib?

Thanks a lot in advance!

Regards

Heinrich C. Kuhn
_________________________________________
Depending on what statistics you need and whom you report them to, you may
want to review the guidelines for statistical measures from the ICOLC:

http://www.library.yale.edu/consortia/webstats.html

Thomas Dowling
___________________________________________
My answers refer only to the Library's pages on the Intranet and Internet
sites of the Ontario Legislative Assembly.  Most of our web-based services
are available only on our Intranet since we do not serve the general public.

Are you reporting hits on specific pages or overall hits
to the site? 
We do both.  We can track the actual hits on any or all of our pages.  We
track key pages and note total hits monthly.

 How are you reporting in-house vs. remote hits?  
We report in house and external hits for each of the tracked pages on our
intranet.  We do not bother to check inhouse hits on our Internet site since
our clients use our services mostly through our Intranet. We are interested
in remote Internet hits, however, since we disseminate some research reports
that way.

Do you report trails within your site or referring sites?  
No

Are there standards for reporting statistics?
I don't know exactly what you mean here.
We have a standard (standard created by us) brief form and a standard
detailed form.  The brief form is done on a spread sheet with month to month
and year to year statistics.  Its data goes into the annual statistical data
for the Library.  The detailed form is used by Intranet staff to assess use
of services and to plan for future services. We track the "home pages" of
our services as well as selected documents within the intranet to assess
use.

I have seen only one article on bench marking web usage statistics.
I would love to see that article.  Could you send me the citation?
We are looking at bench marks for use on our pages:  realizing that certain
of our services will be very heavily used and others not. We only have about
10 months of data so far and have set tentative bench marks only. Will
verify those benchmarks in the fall.

Does anyone know of other articles out there?
No

Susanne Hynes
____________________________________________
Analog, an excellent (and free) log analysis tool, is reportedly used by 25%
of the webmasters out there.  I would highly recommend this product.  It is
very flexible, runs on most platforms, and can create reports in HTML,
ASCII, or CSV (for further spreadsheet tweaking if that is actually
necessary).  Analog's documentation has the following information about they
types of statistics that can and cannot be calculated.

>From one of the Analog documentation
pages ------------------<http://www.statslab.cam.ac.uk/~sret1/analog/docs/we
bworks.html>:
.. what happens when somebody connects to your web site, and what
statistics you can and can't calculate. There is a lot of confusion about
this. It's not helped by statistics programs which claim to calculate things
which cannot really be calculated, only estimated, with varying degrees of
accuracy. The simple fact is that certain data which we would like to know
and which we expect to know are simply not available.

.. I can recommend four excellent articles about this subject: Interpreting
WWW Statistics by Doug Linder
<http://gopher.nara.gov:70/0h/what/stats/webanal.html>; Making Sense of Web
Usage Statistics by Dana Noonan<http://www.piperinfo.com/pl01/usage.html>;
Getting Real about Usage Statistics by Tim Stehle
<http://www.wprc.com/wpl/stats.html>; and, the most negative of all, Why Web
Usage Statistics are (Worse Than) Meaningless by Jeff Goldberg
<http://www.cranfield.ac.uk/stats/>.
--------------------------- END|

I hope that addresses part of your question.  Other answers are interspersed
below...

----- Original Message -----

> Hi!  I am curious as to how you report your statistics gathered from
> your web site?

I use analog to generate reports about various sections of our web site. As
part of the reporting process the monthly reports are sent to the primary
document masters of those sections.

> Are you reporting hits on specific pages or overall hits
> to the site?

both

> How are you reporting in-house vs. remote hits?

At the moment I'm report hits per page and have not created in-house vs.
remote comparisons.

> Do you
> report trails within your site or referring sites?

I don't believe trails within a site can be done reliably (it's not a
feature of Analog so I'm not doing it).  At least one thread I read
suggested that trail reporting is so inaccurate it should not be done.  I'll
leave it up to others to debate the merits of trails.   I do report the
referrer section of my logs.

Hope that helps.
--
John.Litle at Duke.edu
___________________________________________
At 10:58 AM 6/22/99 -0700, John Little wrote:
>  It's not helped by statistics programs which claim to calculate things
>which cannot really be calculated, only estimated, with varying degrees of 
>accuracy. The simple fact is that certain data which we would like to know 
>and which we expect to know are simply not available.

This is true, of course, but also true of almost every other statistic that 
is collected by librarians to "show their worth" or to appease their 
bosses.  Yes, we know pretty well how many books the library owns.  But 
what does that mean?  Not much.  Others such as door counts, reference 
transactions, items circulated, and so forth may have marginal value for 
planning of staffing needs, but certainly don't tell anything about the 
value or utility of the library.

And the same is true of any web stats we obtain, though they may indeed 
help us personally by making our bosses happy with us.

dan