Additionally, our client would be able to implement a single segmentation tool into its core business much more efficiently than two separate tools. The challenge was to therefore decide whether a one tool or two tool solution would be most suitable.
We interviewed an extremely robust global, quantitative sample and then worked with our statistician to apply and test a multi-level segmentation model. The first level of the model segmented our sample using all input variables, based on physician attitudes, behaviour and characteristics, whether indication specific or not. This is the standard segmentation approach, but we then needed to ensure that the physicians we classified in the same segment, thought and behaved similarly to one another towards the different indications. It is important to note that we did not expect individual physician perceptions and behaviour towards the different indications to be the same, but we did expect all physicians from a particular segment to have similar approaches to treating the two different indications. If there was significant differentiation within individual segments based on how they thought about and treated different indications, then we would know that a single segmentation tool would not be feasible.
Our model showed that the relative differences in response across the different indications is similar for ALL segments and independent of segment membership; there was little or no evidence of differences with regards to indication by segment. This lack of statistical significance identified that a combined indication solution was the most appropriate approach – within each individual segment the relative difference in the way physicians think and behave towards indications is similar. The client has rolled out the global segmentation within its business, including adoption from the sales force to help understand individual physician needs and improve these sales interactions.