Google’s average session duration is what it says on the tin: how long users dwell on your website page in question.
The problem is that this includes sessions of 0 second duration. So this value is generally understated.
How the maths works:
Total time of all sessions divided by total sessions (including session where nil time on page is recorded). So if you have 20 sessions toalling 200 seconds, but 19 off the sessions scored 0 seconds, then the average session duration is 10 seconds. Calculation: 200 total seconds / 20 total sessions.
What creates 0 second sessions? Sessions where a user “bounced”. That is, where a user arrived, read your page, and closed the page. More on this under bounce rate.
Thus, the average session duration number provided by Google Analytics is not reliable.
The solution or “work around” here
- Easy: look at a graph which visually displays your user session time
- Hard: try avoid your users “bouncing”
This solves for the “bounced” sessions, as well as outliers on the longer duration (larger value) end of the spectrum. Both of these distort your average value.