Customer Analytics Suite

Challenge

Our client, a medium-sized German manufacturer of technical products, aims to support their sales efforts with data-driven decisions to identify new sales opportunities more quickly. This endeavor is complicated by the high heterogeneity among the company's customers from different industries, who also exhibit very different purchasing behaviors. To address this problem, various data sources need to be combined, and generalizable customer groupings and purchasing patterns must be derived.

Approach

To support sales, we developed an application that uses machine learning to estimate which products should be suggested to customers next and in which product areas they are currently purchasing less than similar customers. These suggestions are derived based on similarities between customers. Initially, we grouped customers with similar use cases, industries, and purchasing behaviors into clusters. For this purpose, we utilized data from the client's CRM system, among other sources.

To condense the large number of influencing factors, we employed a data dimensionality reduction technique before clustering to compress the information. Additionally, we weighted the influencing factors by manually and customer-specifically adjusting the variance.

In the final step of the whitespot analysis, the percentage shares of product categories within a cluster were compared. By analyzing deviations from the average behavior of all customers within a cluster, specific potentials were calculated for each product group.

Result

Based on the tool's results, we helped our client streamline and enhance the efficiency of their sales process: Customers with high revenue potential can be identified much faster and targeted more effectively with tailored sales measures. This includes expanding product categories where customers have already made purchases, as well as making targeted suggestions for additional new products for the customers. The manual weighting of features was particularly helpful in ensuring that the clustering met our client's expectations.