If you track this topic in the list archives over the last eight years you will see that most of what you mention has been covered before. In particular, there is always a management class that will pay someone to tell them what they want to see. And there will always be companies willing to take $100k to give you numbers that make you happy -- whether or not they have any bearing on reality. There are lots of other issues in there too. It's an interesting discussion.
I actually agree with you on your point about session tracking. I think most web sites these days are built around users sessions (at least all e-commerce is). All the back-end tools include support for it (PHP, ASP, JSP, J2EE, ASP.NET, etc.). In this respect Analog's domain box has not really kept up with the times. However, I believe that the user report is not very useful as is. Sure the user report is interesting if you want to look at session ID's and how many requests and page requests and bytes each "user" had, but that doesn't really tell you anything interesting. I think the only really intriguing bit of information from the User Report, when used to track sessions, is the total number of sessions. This can be a useful metric -- especially when tracked over time. Another "meta" metric could be useful too: average number of pages requested. This is not too hard to post-process. And any of the rest of the "interesting" values either conflict with http://analog.cx/docs/faq.html#faq128 or are cannot be counted in a strictly accurate reading and thus could produce reports that appear to say one thing and really say another -- and this is what Analog has always tried not to do. -- Jeremy Wadsack Wadsack-Allen Digital Group [EMAIL PROTECTED] <[EMAIL PROTECTED]> (Monday, February 23, 2004 11:25 PM): > Hi, > Given the recent spurt of discussion on user tracking, I re-read the "how the web > works" bit. While it is true and I agree 100% with it, I find it a bit one-sided. > It's working from the assumption that you absolutely do not want to implement > mandatory login or cookies. This, IMHO, is not true anymore. Login is one thing, but > even mandatory cookies are > nothing out of the ordinary. The web has become more exclusive -- many sites are > designed on the "works with IE, job done" principle, with absolutely no > consideration for people with other > browsers, people who don't want cookies or Javasript etc. Perhaps sad, but true > nonetheless. > E.g. many Java web application base their functionality on storing variables in the > server side session context object. The session id is stored in a cookie (or, > optionally a parameter in the > URL). Without a valid session id the application just will not work, and booting > such a request to the front page is absolutely correct behavior. Thus, you WILL have > a unique session id and even a > convenient starting point for click-tracking. > And this is actually pretty trivial to implement in any J2EE web container. > Yes, there are still "black holes" in the information, but this model is > substantially better (in the user tracking sense) than the classic "random access" > model. > I find Analog a bit schizoid in this sense -- first it provides the USER reports, > barely mentioned in the documentation, then it dedicates a full page to explaining > why the USER report is > useless... (it isn't, but that's the impression one could easily get) > Sigh... I really like Analog. I've been using it for log analysis on a very big > site, with success. But in the future, the management has chosen to go with HitList, > for two primary reasons: > 1) It's a commercial product. I guess this is based on the "we paid them, we can > blame them" principle, even though 99% of software is sold without warranty > 2) It gives the numbers (visitors etc.) the management wants to see -- whether they > mean anything or not... > It does produce rather pretty reports, I gotta give it that. +------------------------------------------------------------------------ | TO UNSUBSCRIBE from this list: | http://lists.isite.net/listgate/analog-help/unsubscribe.html | | Digest version: http://lists.isite.net/listgate/analog-help-digest/ | Usenet version: news://news.gmane.org/gmane.comp.web.analog.general | List archives: http://www.analog.cx/docs/mailing.html#listarchives +------------------------------------------------------------------------
