From: Lafayette, LA USA
I have recently been evaluating reporting suites for ISA to determine employee web usage. I have tried WebTrends, BTLogAnalyzer, and WebSpy. When analyzing the same log file, I end up with 3 different results when trying to determine the amount of time an employee has spent using the WWW.
Does anyone have any suggestion of how to determine which one is correct or have any general advice?
It all depends on their browse time sensitivity. If you take a chunk of browse time and say that any activity within a 3 minute period constitutes the same session...and then take the same chunk of time but use 5 minutes, you will get different totals. There should be a place to adjust this number in the applications you are using. Everybody has a different idea as to how much idle time constitutes a new session.
We use Burstek and their time is how long it took to download a page, which is the only true measure that can be made. Anything else is just a guess.
So if someone shows one hour of time, it actually took one hour to retrieve the pages. If you figure it takes about one second to retrieve a page and a person views it for one minute, that one hour of Burstek time translates to about sixty hours of actual viewing time.
But this is still a guess. The only real way to get a good guess is to load the raw log files into Excel, sort on time and date and then evaluate the times individually.