Hi,
I'm not sure all web statistics are as useless as this document suggests,
working for MCB University Press, we spent a lot of time on our usage
statistics package and are confident that we can give an accurate amount of
'sessions' use. This means we can assign some 'assessment' of the hits in
total.
The way we have done this is, is to assign ID's to each session. We can
calculate how many hits occurred during each session on each ID.
Proxy servers are a major headache for us, our webstats do not show accurate
information from the end user only from the proxy server. Our response to
this has been to ask all the major proxy servers to add our URL's to
'exception tables' so that the user is diverted directly to the resource.
This is obviously not an ideal situation, but it is one that satisfies the
need for accurate access control and the usage statistics.
If anyone requires a sample of our 'basic' usage statistics package it is
available free to subscribers and an example is held at:
http://www.emerald-library.com/EMR/help/EMRusage.html
Kind regards,
Dominic
--
Dominic Martinez
MCB Customer Support Services Manager
mailto:[log in to unmask]
tel: +44 01274 785139
fax: +44 01274 785204
-----Original Message-----
From: Roddy MacLeod [mailto:[log in to unmask]]
Sent: Wednesday, November 03, 1999 10:53 AM
To: [log in to unmask]
Subject: Re: usage stats and caches
elib,
There's some interesting info in the paper suggested by Simon, along the
lines of "Why web usage statistics are (worse than) meaningless". Some
of it I was aware of, and some I'd forgotten.
So - assuming one cannot realistically use usage logs to gauge
page views, does anyone have any suggestions for how to
persuade potential advertisers that its worth their while parting with
their money?
Roddy
> Roddy,
>
> Take a look at a paper written by Jeff Goldberg from
> Cranfield at:
>
> http://www.cranfield.ac.uk/docs/stats/
>
> Rgds
>
> Simon
>
> On Wed, 3 Nov 1999 09:40:47 -0000 Roddy MacLeod
> <[log in to unmask]> wrote:
>
> > elib,
> >
> > Like other services, I imagine, we at EEVL collect stats of hits to
> > our Web pages. We have accurate stats of *searches* made, due
> > to the fact that these are dynamically generated, but other stats
> > are, I believe, not accurate due to caching at various levels. Is
> > there any simple theory for the effects of caches on hits to home-
> > pages and other static pages?
> >
> >
> > Roddy MacLeod
> > Senior Faculty Librarian
> > Heriot-Watt University
> > Edinburgh EH14 4AS
> > United Kingdom
> > Email: [log in to unmask]
> > Phone: (0131) 451 3576
> > Fax: (0131) 451 3164
> > Web: http://www.hw.ac.uk/libWWW/libram/roddy.html
>
> ----------------------
> Simon J. Bevan
> Information Systems Manager
> Kings Norton Library
> Cranfield University
> Cranfield
> Beds.
> MK43 0AL
>
> [log in to unmask]
>
> Tel: (+44) (0) 1234 754445
> Fax: (+44) (0) 1234 752391
>
>
Roddy MacLeod
Senior Faculty Librarian
Heriot-Watt University
Edinburgh EH14 4AS
United Kingdom
Email: [log in to unmask]
Phone: (0131) 451 3576
Fax: (0131) 451 3164
Web: http://www.hw.ac.uk/libWWW/libram/roddy.html
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|