For financial services firms, acquiring new customers is more costly than retaining them. Therefore, retaining existing customers is one of the biggest challenges. Customer churn is everywhere. In unsecured lending, customers cancel credit cards or personal loans, or there’s silent attrition in the form of a slow decrease in customer card spend. In secured lending, mortgages face churn in the form of a loan transfer to other lenders, partial or full payment of loans and loan closure. Customers also can close their bank accounts, resulting in the loss of potentially cheap sources of funds, or they cancel their life or general insurance policies, resulting in the loss of potential future premiums.
Moreover, customers often have multiple relationships across different services (banking, loans, insurance, etc.), and a churn in one relationship can trigger churn in other relationships.
Churn is an issue internally, as well. Just like silent attrition in credit cards, attrition of sales agents and decline in sales performance also can be a challenge.
To help solve the churn problem, firms should take a data-driven approach by developing churn prediction models. These models identify a churn event using transactional, demographic, product and pricing data.
However, ZS research has found that many companies struggle with four key components of churn prediction models:
- Developing a definition of churn that’s easy to understand and makes business sense
- Gathering data—and going beyond what’s normally available—specifically for churn modeling
- Applying sophisticated machine-learning algorithms to work with big data
- Developing infrastructure to automate the prediction results and create actionable insights, and for helping business managers use the new process and move away from making gut-based decisions
Mitigating these challenges, and getting the most from your churn management processes, requires a concerted effort. In my new white paper, “Churn Prediction: Five Success Stories,” I share a model that helped predict decline in sales agent performance and engagement, along with methods to solve the churn problem holistically. In addition, I also share four other case studies that showcase how companies solve churn prediction challenges using data science techniques, and these varied examples show financial services firms how to predict and prevent churn, and address the four challenges of churn prediction.
Customer and employee churn are invariably costly problems. These problems can be solved by cleverly crafting a solution to the churn problem and using advanced data science techniques on available data. It’s worth it to invest in a data-driven approach that heads them off at the pass.