Introduction

A Temenos Analytics Customer Lifetime Value (CLV) Predictive Scenario is a forecasting model used to estimate the average profit generated by a customer over a period of time (a time horizon). Such estimations are of great business value for creating targeted and efficient marketing campaigns. Customer Retention campaigns are a good example – it is always more efficient to try to retain not all customers that are predicted to churn (e.g. by using A Temenos Analytics Customer Attrition predictive scenario), but only those that are reasonably expected to generate the most profit within certain time horizon.

The Time Horizon is a major consideration when estimating lifetime value. Perhaps the most natural perception of Lifetime Value is the one that spans the entire lifetime of a person. Although very intuitive such perception is not necessarily the best way to look at CLV in a business setting. More often than not, businesses care for much shorter time horizons, e.g. several years rather than decades. Therefore, the Time Horizon should be determined not so much based on life expectancy but rather on things like average customer tenure and/or business targets and/or strategic goals of the bank.

The Temenos Analytics CLV Predictive Scenario comes in two editions – Standard and Advanced.

The Standard edition is developed to accommodate new implementations of Temenos Analytics where the amount of data in Warehouse database is limited. This Edition needs only one month of data to operate.

The Advanced edition should be used in all situations where there is already 13 months of data in the Warehouse database. This Edition provides greater accuracy and model exploration affinity and should replace the Standard Edition as soon as enough data is accumulated in the warehouse.

Default Model Name1 or other unique name specified at installation time	Description
CLV_Segments (Advanced Edition, step 1)	This model implements the first step in the Advanced edition scenario. It builds segments of Customers \ Product Classifications based on their underlying profitability drivers – balances, interest rates, fees and costs, by using the Decision Trees algorithm.
CLV_Calc (Advanced Edition, step 2)	This model implements the second step in the Advanced edition scenario. It takes the Customer \ Product segments from the first step and builds Customer segments that share similar mix of products and profitability drivers. A first order Markov’s chain algorithm is used to build a transition matrix that contains the probabilities of transitioning customers from one Customer segment to another. The transition matrix is used to estimate CLV over the desired Time Horizon.
CLV_Standard (Standard Edition)	This model implements the Standard edition scenario. Because of the limited historical data in the warehouse, there is no segmentation step. All customers are considered as belonging to a single ‘default’ segment and the CLV is estimated by using the average profitability of customers at different ages.