[WEB4LIB] RE: website stats analysis
Robert Sullivan
rsullivan at sals.edu
Fri Aug 6 15:47:37 EDT 2004
> Actually, differentiating by IP address depends on your institution's
> network configuration. At our institution, all computers are assigned
> dynamic IP addresses whenever a user "logs on" to the Internet. While
> many of the IP address ranges are grouped geographically, in the
> physical library itself, the client and staff computers are not
> separated. We have, therefore, been unable, so far, to remove
staff-use
> data from our logs.
>
> I must question, however, the value of removing such data, because
many
> of our staff are working either with or for clients, either directly
or
> indirectly. Furthermore, shouldn't the staff be considered clients,
as
> well? Are they any more likely to browse out-of-scope sites than
> students, faculty or other staff?
That's been my philosophy too. I read my Web site logs into a Visual
FoxPro database and remove any hits from a long list of robots and
spiders, then I look for anomalous use patterns which suggest the same.
After some careful pruning (I'd say about 70-80% of our log entries are
extraneous) eventually we end up with a fairly accurate count of page
views. I prefer grouping them into categories rather than worrying
about specific pages; it seems to give me a better sense of what our
visitors are using.
I don't try for an actual count of Web site visitors because I'm more
interested in what's used than in who's using it. On the other hand,
for our database subscriptions sometimes "visitors/users/sessions" and
perhaps "searches" may be the only useful count you can get from the
database vendors, especially if you're trying to compare very different
kinds of services. A hit count is essentially useless IMHO in that
situation. What you really want to know is how many people used that
subscription for which you paid $x thousand.
Your mileage may vary.
Bob Sullivan <rsullivan at sals.edu>
Schenectady County Public Library (NY) <http://www.scpl.org/>
Schenectady Digital History Archive
<http://www.schenectadyhistory.org/>
More information about the Web4lib
mailing list