Part 2: How Do I Better Market to People Outside my Customer Base?

New Call-to-action

Part 2: How Do I Better Market to People Outside my Customer Base?

 In part one of this two-part series, we explored targeted marketing within an institution’s existing customer base. In part two we address the second aspect of Customer Analytics: marketing to people outside of your existing customer base. In this scenario, we will introduce a new type of customer analytics-- advanced segmentation.

Advanced Segmentation

If you recall our discussion of wallet share from Part I a of this series, you will remember the power of being able to segment customers and see the overall distribution of wallet share both within and outside an institution. Here we will present a more algorithmic way of grouping people for targeting those outside of our institution.

 As is common in all Customer Analytics, we first want to organize our customers along with all individuals from the credit bureau data. However, we are not just organizing by borrower age or income as we’ve seen previously. This time we are looking at multiple characteristics of a borrower’s attributes which allows us to use our algorithm and group people based on similar demographic traits, product history, or preferences.

 Clustering algorithms are powerful tools in this analysis. Once we have selected which borrower characteristics will be used in determining groups, we calculate the “distance” between borrowers and use that distance measure to cluster individuals into segments. Distance in this scenario involves a mathematical metric used to quantify individuals by their similarities and differences. In the plots below we see this in action. On the left, we see a large spread of borrower data without much differentiation. Once our clustering algorithm is put to work, we create the plot on the right. The right plot divides the left plot into three distinct groups, which were separated based on the chosen variables and distance metric.


 Once we determine the number of clusters we wish to analyze, we can interpret these clusters and begin to create “personas” for each segment.

 Here’s the goal: when we group similar people together in a particular segment, we can expect similar responses from them. Each individual segment then will respond differently to different product and rate offerings. As you can guess, these segments make our marketing efforts much easier and tailor-fit as we know which segments to target and how we can expect the borrowers to generally respond. Added predictability is a bonus in any marketing effort. 

Keaton Baughan

Product Manager