Customer segmentation is an essential and valuable exercise for marketers in the financial services industry. The many benefits of a customer segmentation program include the ability to target campaigns and products to the most profitable audiences, increase response rates, lower acquisition costs and increase retention and cross-sell rates. Segmentation can also help banks compete with non-traditional entrants to the financial services market, providing a competitive advantage by leveraging the data they have about their customers.
Many financial institutions use segmentation trees to group customers with common attributes or behaviors. Segmentation trees are a simple way to define customer segments with similar values through a set of rules. They are popular because they are easy to use, and easy to understand. However, they also have some disadvantages.
Often marketers approach the segmentation effort with some assumptions regarding potential opportunities. For instance, marketers may assume customers with the greatest appetite for new credit products live in wealthier zip codes, so they start examining that groups’ purchasing behavior while ignoring middle tier zip codes where there may be more opportunity.
This type of segmentation exercise may result in some missed opportunities —in other words, it may not be able to find “hidden” patterns. In addition, the segmentation output may have limited usage for other business initiatives.
Big Data and Banks
Financial institutions have access to significant amounts of customer data. From browsing patterns on banking website and apps, to demographic information, to product and transactional data such as checking activity, deposit balances, loan usage and mortgage rates, banks are collecting a massive amount of information on their customers’ habits, attributes and behaviors. Segmentation trees can’t necessarily leverage this data effectively if the marketer performing the segmentation doesn’t know or understand all the information they have at their fingertips – a common scenario, given that data collection may happen in silos. Thus, banks may not be aware they are missing key data points that can help them predict certain customer behaviors when running segmentation trees.
R-based Mathematical Segmentation: Another Approach
Let’s look at a different method of segmentation that removes some of these obstacles: a statistical approach using k-means clustering in the R programming language. K-means clustering is a sophisticated methodology of mapping a large set of values to a smaller, countable set, used in data mining. While segmentation trees are based on logic (If A, then B), k-means clustering is an entirely mathematical method of classifying data points into clusters using an algorithm. With k-means clustering, there is no way to manipulate the groupings or apply bias or assumptions.
Some of the benefits of this approach include:
- The ability to analyze thousands or millions of data points: there’s no limit to the amount of columns or rows of data that can be leveraged.
- Easily slice and dice customer data into infinite number of groups: the segmentation allows you to easily re-run data into sub-segments to focus on particular attributes of a group.
- Correlate and predict behavior more accurately: k-means clustering can help uncover relationships and shared attributes between customers that you might not have known existed. For instance you might learn that a large portion of low profitability customers have significant potential – driving marketers to reorient their retention and cross-sell targeting.
- Provide better, mathematically validated data to back up decisions.
Like any other method of segmentation, k-means clustering does not provide all the answers – the savvy marketer will still need to use their expertise to understand what the relationships between the clusters mean for their business and how best to target those groups. However, the insights gained from using this approach and the related benefits – such as reduced time and expense to go to market— can generate incredible payback and bottom line benefits that make the upfront investment worthwhile.