Usage statistics

[Daniel Feenberg <feenberg@nber.org>]

>From time to time we get requests from library subscribers to our online
service for statistical information about their patrons use of our
service. At least one librarian asked that only her PC be authorized to
receive documents, so that she could keep a usage count herself.

Before today, we have been putting off requests for statistics but a few
weeks ago I saw a suggestion in this list that publishers place the logo
of the subscriber on the page viewed by the library patron, so that the
patron would remember who was paying for the service. I have recently
realized that we could use this logo URL to solve the statistics problem
also.

The logo served by us is just an href to a URL on the subscriber's own web
site. So whenever a patron views a page, the library logo is obtained from
the libraries own web server.  The library could use whatever web log
analysis tools they already have to gather whatever statistics they want
just by counting hits on that URL. Our prospective customer accepted this
plan, and we are trying it out now. I don't know if they consider it a
success yet, but it seemed worth sharing the idea with this list in any
case.

If the library is already keeping track of some local content from web
server logs, it gets the statistics for the outside content in the same
format and with very little effort.

There are some technical details to keep caching from distorting the
statistics, but overall it seems like a good solution to me. I would like
to thank the person who posted the original suggestion, but I did not save
the posting.

The web log will include the IP address of the patron viewing the
document, and the URL of the page displayed to the library patron.


Daniel Feenberg
feenberg@nber.org