One of the things I like most about my job is the variety of work I do and the range of organisations I work with. Sometimes I help established agencies deliver a brief, usually because they are temporarily over-stretched or in need of a particular expertise. It’s always interesting . This week I’ve been helping a research agency with the technical aspects of a survey-based segmentation. It’s been fun revisiting a category I first worked on over 20 years ago. I know, I can’t believe it either.
To be honest, the client did all the hard work and I did the fun stuff. They designed a good online survey which, for the uninitated, is harder than you might think. They jumped through all the hoops required to get a good survey completed by good respondents, again, something that’s a lot harder than you might think. They processed all the data and passed it on to me and I got to do my favourite thing: building segmentations. Actually, I built two so the end client could choose the option that made most sense for them.
I won’t bore you with how I did it except to say I used an old-fashioned but reliable technique called factor analysis to identify groups of people who had simiar attitudes to food and a particular type of cooking sauce all based on how they answered survey questions . There are always lots of different approaches and there’s a fair bit of fiddling around taking some questions out and putting more in until I have groupings that meet my friend Julie’s segmentation rules (I wrote another blog on that). Julie’s rules are that segmentations must be:
- Undertstandable (the differences between the groups must be obvious)
- Actionable (the results have the potential to drive a change)
- Sizeable (neither too big nor too small)
Once all this is done and the end client agrees on which option they want, I tidy up the data and send it back to my client who does the hard work of using examining the data and bringing each segment to life for their client.
Always nice to get some feedback from the client who said:
