Introduction

A Temenos Analytics Customer Attrition Predictive Scenario is a predictive scenario, implemented by using a decision trees algorithm. It allows financial institutions to determine which active customers are likely to attrite (churn) during the prediction time horizon - usually 60 to 90 days, based on their behaviour in the past.

The ability to predict customer attrition (churn) is an important part of any company management. The key point is that new customers are expensive to acquire and generally generate less revenue in the near term than established customers. After the initial period of exponential growth of the business has been left behind, churn modelling could be successfully applied to focus the retention efforts on high risk customers who might leave without the extra incentive.

A combination of rule-based data transformation and a data mining models is used in a stepped approach to predict the risk of customer attrition. The first step involves identifying the latest churners by using a Packaged Transformation that implements configurable rules. The latter is used as one of the inputs for the second step.

Once the list of the churners has been generated, a decision trees algorithm is used to build a classifier that predicts churn.

In essence the step 2 model is trying to sniff behaviour of disengagement, such as less deposits, less transactions, less products, which eventually leads to attrition. This disengagement happens over a period of a few weeks to a few months and can express itself in increasing/decreasing the values of total balances, number of products, transactions, credit transactions, pre-authorized debit transactions, i.e. recurring expenses, customer initiated debit transactions, i.e. day-to-day expenses, and so on.