Computing Average Website Visit Length

[Gorilla]

Is there a commonly accepted way of including one-page visits when computing average website visit lengths? Are they excluded, perhaps reported separately?

When computing the length of a visit, does one simply subtract the time for the first access from the time when the last access occured? This does not take into account the time spent viewing the last page loaded.

[Receptional]

I have not heard of an agreed standard for this.

You could analyse the average time between page views for all visitors on all pages and then call this an average for every person X number of pages visited, although I suspect the landing page has different averages to other pages on the site.

Of course, doing this calculation will dramtically improve the total time people spend on the site - because people visiting three pages with 30 second intervals take up 1 minute in the logs, but the average is 30 seconds to read a page - so potentially all tracking systems are "forgetting the time to read the last page the user reads"... Interesting thought.

Dixon.

[webnerd]

A similar but somewhat opposite type problem is the case of webTV ips which use one ip for all their visitors to their system.
The stay length for the ip can sometimes be up 24 hours because there is another of their tv users is accessing another page of my web site in less then every 15 minutes.
Since I have found about 20 seperate webtvs varying from x to 24 hours this tends to skew the average visitor length of time also not to speak of the actual number visitors.
I think log file analysis reports are at best a "guessimation" of the traffic to a given website.