A glimpse of what our speakers are presenting at the 2022 Human Insights Conference!
What’s the biggest driver? Why it’s not just size that matters in driver analysis but what you do with it.
Are you running a brand centric driver model anytime soon? Will you be making big decisions off that model?
Here is something you might want to consider, and you don’t need a stats background to get your head around it!
As insights professionals we regularly use driver analysis to answer some of the biggest strategic questions, the results of which often dictate where an organisation will invest to optimize growth and improve returns.
So, it is critical we design and run statistically strong driver models, but it is equally critical we interpret the model outputs correctly.
When it comes to running a driver model most of the debate has centered on which analysis to undertake to ensure statistical strength, given the success of a driver model is highly dependent on managing multi-collinearity and missing data.
But there is little debate on how we interpret the model outputs.
A driver model identifies the attributes that best explain the differences between people who score low on a dependent variable (what we are trying to predict – such as NPS, customer satisfaction, brand preference etc.) vs those who score high – for simplicity let’s call these happy and unhappy customers.
In a driver analysis output the bigger the coefficient for an attribute, the bigger the difference in the attribute ratings between happy and unhappy customers. This coefficient is often reported as the ‘importance’ score.
Typically, when interpreting a driver model, the attributes with the largest ‘importance’ scores are regarded as the ones needing the greatest attention and investment. Because they explain the most difference between happy and unhappy customers.
We want to introduce a new way of thinking when interpreting these attribute coefficients that are reported as ‘importance’ scores.
Remember the classic saying “correlation is not causation”, well, here is a new one, “importance is not influence”.
Here are three things to consider about how influential that importance score is
- In brand centric models, not all attributes with the biggest importance scores are the most influential in driving brand choice. In fact, many of the ‘fundamental pillars’ of brand choice typically receive smaller importance scores in a driver model. That is because both happy and unhappy customers agree the brand can deliver on that attribute. Suddenly, that attribute with a low ‘importance’ score in your driver model report turns out to be one of the most influential reasons customers choose your brand.
- A driver analysis is conducted at a point in time, it considers both the current number of customers exposed to an attribute as well as your brands current performance. But what is important today may not be your biggest differentiator in the future. You need to be able to identify which attributes have potential to drive future competitive advantage.
- Attributes with the largest importance scores could be your brands greatest strengths or weaknesses – you need to know if you should be fixing or promoting these attributes.
We will present a new way of interpreting brand centric driver models that is not laden with jargon and empowers the user to think more strategically about where to invest.
Presenter: Lyndall Spooner, Fifth Dimension Research and Consulting