This week I’ve been working on one of my favourite market research techniques: Key Drivers Analysis.
A key driver analysis is a useful way of understanding which components are important for a customer when making a specific decision. For example, when choosing a product, like a box of tea bags does brand matter more than size? Or, when you’ve eaten in a restaurant does atmosphere matter more than food choice when it comes to overall satisfaction?
This week I was working on a survey for a news publication, one of the things we really want to understand is what features influence whether a customer will renew their subscription.
The questionnaire asked subscribers whether or not they were likely to renew’ and they also asked a series of questions about the features of the publication from overall quality of journalism to areas of interest and preferred format.
Now, before I explain the steps I took I need to say that, despite being evangelical about Q Research when it comes to Key Drivers (and other complicated analytical techniques) I prefer to use SPSS simply because I can look at the workings and that helps me interpret the findings (and question them).
So, this is how I do a Key Driver Analysis:
- I need to identify both sets of variables. First, the ‘outcome variable‘ (usually self explanatory and something like intent to purchase, satisfaction, likelihood to recommend) in this case the outcome variable was ‘Likelihood to recommend’. Secondly the ‘driver variables’ these are all the variables that influence the outcome.
- I want to make sure the final result is sensible so I run a correspondence map (or principal component analysis) which provides a visual to identifies the relationship between each of the variables. This tells me which variables are outliers and I tend to (but don’t always) remove these outliers from my final list of driver variables.
- Then I run a regression analysis which does two things: it gives me a mean score for each of the driver variables and standardised co-efficient (which is shows us the importance of each driver variable in determining the outcome variable).
- Then I standardise the mean and the (already standardised) co-efficient. What this means is that both variables are forced into the same range and structure which means they can be plotted nicely on a scatter chart
- Once all this has been done I can look at the output. Anything with a standardised mean of more than zero is rated highly by respondents, anything with a standardised co-efficient of more than zero is ‘important’ in determining the outcome
And what we end up with looks something like this:
The green variables are the ones that are performing well and drive importance. The red are the variables that need work they’re important in driving likelihood to renew BUT they aren’t scored that highly – these are the areas that need a consistent, strong focus, they need to be improved.
If you think this type of analysis would help you make better decisions or inform where you should be focussing your efforts get in touch!
Thanks as always to David Dipple, my friend and mentor who gifted me with these skills more than five years ago and continues to be the person I contact when I’m just not sure how to make sense of the numbers!