Question

Topic: Customer Behavior

Multi-domain Segmentation

Posted by Anonymous on 250 Points
Could any one please help me to understand on how to perform a Multi-Domain Segmentation (MDS)?
Let's say we have consumer survey data for an FMCG Brand, and we have details for following Domain (each domain will have 20 questions)
1) Demographic
2) Usage
3) Attitude
4) Loyalty
5) Lifestyle

While it is easy to do a segmentation based on only one of the domain (using k-means/two step cluster analysis). how to do a multi domain segmentation using all above mentioned domains? what statistical methodologies should be used to perform the same?

Many Thanks
To continue reading this question and the solution, sign up ... it's free!

RESPONSES

  • Posted by Moriarty on Accepted
    Pradip

    [rant] Fair Warning: Analysis is not my strength. I did Natural Sciences at Uni* and started with biology because I love plants. I couldn't stand all the ruddy statistics. All that dumb number crunching and not a moment to enjoy the Botanic Gardens! I love plants, statistics reduces them to a pile of meaningless numbers. The results they got were dry and dead. All the beauty I adored had vanished into monochrome spreadsheets.
    [/rant]

    So what has this to do with domains? Well take it from the Moriarty perspective. That is to say, to kick the ball way out of the field and start playing with it there.

    First point - who is making this data that you want to analyze? Because as soon as you switch your perspective you see another dimension altogether. Looking at the people who make your data will give you a whole new dimension on what the data says.

    Understanding what these people feel is what you are dealing with here. Because feeling is what motivates people.

    So: an apology, because this hasn't helped you one iota, has it?

    Only when you do your statistical analysis, remember that these aren't just numbers. Someone somewhere has done something, and somebody was watching. That is when the numbers really become powerful. That is when you can add the life back in.

    Over to the Number Crunching Crew now.

    To your success, Moriarty.

    (* Everyone says now "You couldn't decide, could you" - and they are right!)
  • Posted by mgoodman on Accepted
    So let's start with the objective: Why are you doing this? How will you use the results?

    Your response will directly impact the methodology.

Post a Comment