Comments on: A sample of how my visitor engagement index drives insights http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html Eric T. Peterson's Web Analytics Demystified weblog, since 2005! Sat, 11 Oct 2008 07:54:11 +0000 http://wordpress.org/?v=2.3.2 By: Web Analytics Demystified » Blog Archive » How to measure visitor engagement, redux http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-40362 Web Analytics Demystified » Blog Archive » How to measure visitor engagement, redux Tue, 23 Oct 2007 17:12:33 +0000 http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-40362 [...] I then went and wrote over a dozen posts, publishing feedback from some incredibly bright people and demonstrating the utility of a well-defined measure for engagement. Since that time, however, some have questioned the value of such a metric and thusly prompted me to update and publish the following calculation for visitor engagement: [...] […] I then went and wrote over a dozen posts, publishing feedback from some incredibly bright people and demonstrating the utility of a well-defined measure for engagement. Since that time, however, some have questioned the value of such a metric and thusly prompted me to update and publish the following calculation for visitor engagement: […]

]]> By: Web Analytics Demystified » Blog Archive » Is engagement an excuse? http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-37465 Web Analytics Demystified » Blog Archive » Is engagement an excuse? Sun, 14 Oct 2007 06:50:09 +0000 http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-37465 [...] Perhaps ironically, in an interview with Eric Enge from February of this year, Enge asked Kaushilk about my long series of posts on measuring engagement (emphasis mine) Eric Enge: Another thing I read about recently was Eric Peterson’s notion of an engagement metric. Can you comment on that? [...] […] Perhaps ironically, in an interview with Eric Enge from February of this year, Enge asked Kaushilk about my long series of posts on measuring engagement (emphasis mine) Eric Enge: Another thing I read about recently was Eric Peterson’s notion of an engagement metric. Can you comment on that? […]

]]> By: Anonymous http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-181 Anonymous Mon, 29 Jan 2007 23:30:00 +0000 http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-181 Again - shoot freely. :-)<br /><br />I'm not sure I'd agree about the bias removal, but emphatically agree with the complexity reducer.<br />I'd be prepared to hang the entire case/argument on reducing complexity alone! :-)<br /><br />My for having a fudge factor bias tho? I've done a lot in IT Security Risk Analysis and that's fudgey left, right and centre.<br />If only because you're dealing with people's perceptions of what's important vs what may actually be a real threat. Which is also *my* perception. :-)<br />It's one of the reasons why it can be useful to get a more or less dispassionate outsider to do the work.<br /><br />It *may* be useful to perhaps chase down some risk analysis work/texts and see if there's any scope for pulling some of their ideas into this engagement concept????<br /><br />The gist being that you take a well known/documented series of issues, and rate each - high, medium, low and that when munched together fudges the various parts to give a Number.<br /><br />I'd be, at least curious, to see how your graphs change if you rated your 5 (?) data points.<br />Say a "2" for a High, 0.8 for a Med, and 0.1 for a low, or vary around that a bit.<br /><br />The idea being: eg "blogging". You and wife have child #12. ;-) You no longer have time for blogging twice a day, and have to cut back to once a week. So what was, say, a High, becomes a Low.<br />The ratio extracted from "Blogging" is now devalued, so what is the possible outcome? It almost becomes a predictive tool in that respect. Trying to guess what may happen if you shuffle priorities around.<br /><br />Which is where the Risk Analysis comes in - you're trying to predict the future on past behaviour and knowns, to best balance various resources to get the best bang for buck.<br /><br />In a perfect world, Web Analytics would be showered with $$. Reality, it probably won't, so we need ways of making it easier to juggle limited resources around.<br /><br />Or am I dribbling. Again. ;-)<br /><br />Cheers!<br />- Steve Again - shoot freely. :-)

I’m not sure I’d agree about the bias removal, but emphatically agree with the complexity reducer.
I’d be prepared to hang the entire case/argument on reducing complexity alone! :-)

My for having a fudge factor bias tho? I’ve done a lot in IT Security Risk Analysis and that’s fudgey left, right and centre.
If only because you’re dealing with people’s perceptions of what’s important vs what may actually be a real threat. Which is also *my* perception. :-)
It’s one of the reasons why it can be useful to get a more or less dispassionate outsider to do the work.

It *may* be useful to perhaps chase down some risk analysis work/texts and see if there’s any scope for pulling some of their ideas into this engagement concept????

The gist being that you take a well known/documented series of issues, and rate each - high, medium, low and that when munched together fudges the various parts to give a Number.

I’d be, at least curious, to see how your graphs change if you rated your 5 (?) data points.
Say a “2″ for a High, 0.8 for a Med, and 0.1 for a low, or vary around that a bit.

The idea being: eg “blogging”. You and wife have child #12. ;-) You no longer have time for blogging twice a day, and have to cut back to once a week. So what was, say, a High, becomes a Low.
The ratio extracted from “Blogging” is now devalued, so what is the possible outcome? It almost becomes a predictive tool in that respect. Trying to guess what may happen if you shuffle priorities around.

Which is where the Risk Analysis comes in - you’re trying to predict the future on past behaviour and knowns, to best balance various resources to get the best bang for buck.

In a perfect world, Web Analytics would be showered with $$. Reality, it probably won’t, so we need ways of making it easier to juggle limited resources around.

Or am I dribbling. Again. ;-)

Cheers!
- Steve

]]> By: Eric http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-182 Eric Mon, 29 Jan 2007 21:05:00 +0000 http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-182 Steve: Yeah, I had looked at applying different weightings to each component metric but given the perceived complexity of the calculation I opted to go for a straight weighting more or less. I think this is the right strategy since it does two things: removes biases derived from assumptions and simplifies the inevitable explanation of what visitor engagement describes.<br /><br />Still, an excellent point! Thanks for the comment. Steve: Yeah, I had looked at applying different weightings to each component metric but given the perceived complexity of the calculation I opted to go for a straight weighting more or less. I think this is the right strategy since it does two things: removes biases derived from assumptions and simplifies the inevitable explanation of what visitor engagement describes.

Still, an excellent point! Thanks for the comment.

]]> By: Anonymous http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-183 Anonymous Mon, 29 Jan 2007 20:52:00 +0000 http://blog.webanalyticsdemystified.com/weblog/2007/01/sample-of-how-my-visitor-engagement.html#comment-183 One of the issues Gary raised was the bias potential.<br />Feel free to shoot me down here, but why not roll in some sort of "Fudge Factor" that deliberately weights the various ratios?<br /><br />The principle being that ratio A is a more valuable ratio - drives book sales, so it's gets a weighting closer to one. Ratio B is useful for long term, but doesn't directly drive book sales so it gets a weighting of, say 0.5. and Hence things pan out nicely.<br /><br />The difficulty is that these fudge factors are prone to abuse, but well, this engagement metric already has a fudge factor of one. :-)<br /><br />You could probably derive the fudge factors from some simple business values? Unsure....<br /><br />???<br />- Steve One of the issues Gary raised was the bias potential.
Feel free to shoot me down here, but why not roll in some sort of “Fudge Factor” that deliberately weights the various ratios?

The principle being that ratio A is a more valuable ratio - drives book sales, so it’s gets a weighting closer to one. Ratio B is useful for long term, but doesn’t directly drive book sales so it gets a weighting of, say 0.5. and Hence things pan out nicely.

The difficulty is that these fudge factors are prone to abuse, but well, this engagement metric already has a fudge factor of one. :-)

You could probably derive the fudge factors from some simple business values? Unsure….

???
- Steve

]]>