"Bryan Ax" <[EMAIL PROTECTED]> wrote.

> I understand that part of the concept - it's just that with huge log
> files,

But that's exactly the point. You don't really care about each of the
50,000 "visitors", you just want an idea of what they're up to. So just
pick 10 or 20 at random, and do individual Analog reports on them. If
they're all pretty similiar, then you've just saved yourself $2500. If
they are all totally different, then you've just saved yourself $2500,
too, because the single "averages" that Webtrends will report will be
pretty meaningless.

Webtrends requires you to use a cookie to actually track users (and it
still manages to come up with soem truely bizarre results that have no
real bearing on reality). If you don't have any sort of "session cookie"
in your logs (ASP creates session cookies, but you may not be logging
them) then Webtrends will try and track visitors by IP address, but it
doesn't bother to tell you that 17 million AOL users share just a handful
of IP addresses, and that any one of those AOL users might use 10 or 15
different IP addresses during a single session, which makes a total
nonsense of any session tracking based on IP addresses.

>I don't know the quickest way to find the "start" of each
> session, or the quickest way to walk thru a whole slew of page requests
> to determine the end of the session (basically, each request will need
> to be compared to the request that came immediately before it and see if
> the time is greater than the time limit). Just use some sort of
> recursive loop to compare files?

No, you need to decide what defines a "user" to you, and then track that
user. If you're logging cookies, and you have some sort of "session
cookie" then configure your LOGFORMAT to treat that cookie as the %u
field, and just use a USERINCLUDE statement to crate an individual report
for a specific user. If you don't have any sort of "visitor" field in your
logs, you can try it with the IP address field, but you'll likely
encounter the difficulties mentioned above.

The key issue is that the numbers that Webtrends will happily deliver to
you don't necessarily mean what they appear to mean on the surface. Analog
doesn't deliver those numbers not because it's hard to do, but because the
numbers aren't reliable, because of the very nature of the way the
Internet is organized.

Aengus

+------------------------------------------------------------------------
|  This is the analog-help mailing list. To unsubscribe from this
|  mailing list, go to
|    http://lists.isite.net/listgate/analog-help/unsubscribe.html
|
|  List archives are available at
|    http://www.mail-archive.com/[email protected]/
|    http://lists.isite.net/listgate/analog-help/archives/
|    http://www.tallylist.com/archives/index.cfm/mlist.7
+------------------------------------------------------------------------

Reply via email to