On measuring engagement
Brooklyn, NY 2007-07-25T00:41:01Z
This was intended as a comment on Registered users? Here's the stat I want to see… Growth of Engagement but apparently San Francisco has gone off the grid and I don't want to lose the thought.
In that post, Charlie O'Donnell noted that many sites report overall user registrations, which tend to be a nice large number, but not user engagement or growth of user engagement.
This is actually pretty easy to track with cookies, but few sites do this, relying instead on a solitary cookie to track both account information for the user as well as the active session.
If they do use session cookies, it's primarily to support the application, eg PHPSESSIONID, and not for audits and measurements.
What I wrote into the typepad bit bucket:
Many sites conflate their session cookies with their account tracking cookies, meaning they only can track when someone "logs in" and resets the cookie. Far better is to split the account cookie from the session cookie, and set the session cookie to expire every hour (well, some unit of time) and then extend the expiration time with each ensuing interaction of the user.
Now, this is sort of the thing that Google have recently come under criticism for (dropping their account cookie time down to a rolling two year period, expiring based on the user's last interaction with the site); however that was specifically an account cookie (containing all the information needed to track the individual) vs. a session oriented cookie (which should contain only the information you need to track an individual session, which you then may or may not want to tie to an individual user depending on your application and ability to gather and analyze the data).
«Your Search Results are not a Suitable Replacement for Navigation | Main | 2007-07-25T1600Z»