CLTV Prediction

Use

As part of CLTV analysis, you can use the CLTV prediction to calculate how many customers are currently in respective lifetime period, how many of these customers are likely to be retained in future, and the profit you can expect to make with their custom. You execute a CLTV prediction on the basis of a CLTV model.

Prerequisites

...

       1.      You need to be assigned to the role Customer Value Analysis ( SAP_BWC_CUSTOMER_VALUE) in SAP BW. You are then authorized to call up the transaction for CLTV modeling from the user menu (by choosing Customer Lifetime Value ® Customer Lifetime Value) and to call up queries for displaying the results.

       2.      You need to have already set up a complete CLTV model during CLTV modeling.

Features

You first need to specify some parameters for the CLTV prediction. These parameters are then used in the calculation.

Settings for the CLTV Prediction

You first specify the CLTV model that is to form the basis of the CLTV prediction.

You can specify whether the prediction is to be executed for the same customers that were used to calculate the CLTV model or whether other customers in that segment are to be considered.

·        If you want to execute the prediction for the customers used in the CLTV model, select the option Data Collection of the CLTV Model (Active Customers). In this case, the system uses for data collection the same data sources as those in the assigned CLTV model. However, only those customers marked in the CLTV model as active customers are taken into account.

·        If you want to execute the prediction for other customers, you need to make new settings for data collection. The data source that you assign for data collection must provide data at the Customer/Segment/Customer Since level. You need to ensure that the segment used corresponds to the segment in the CLTV model. The system considers all customers contained in the data source. You therefore need to make sure that the data source only provides the data for active customers.

If the customer status determined for each customer is to be written to an attribute of each respective customer, select the relevant indicator and specify the attribute. However, the customer status saved is only the status determined the last time the CLTV prediction was performed. If you want to examine changes in customer status over time or use a different method for determining the customer status, you can define your own analysis in the Analysis Process Designer (as exemplified by the template RSAN_PR_TEMPLATES_STATUS).

In the field Number of Prediction Periods, you enter the period in the future for which a prediction is to be made. A prediction period has the same duration as a lifetime period, for which you specified the number of months in the CLTV model.

If the system is unable to automatically determine the customer retention rate and the profit per customer for each lifetime period, you can use the Transfer Last Known Customer Retention Rates and Profits per Customer indicator to specify that the system fill the missing data using the values from the last known lifetime period.

When you have made all the necessary settings, you can execute the CLTV prediction in the background from the Execute tab page. Besides the key date for the prediction (usually the current date or the first day of the current month), you can also specify whether you want to perform data collection concurrently on parallel servers of a server group.

Result of the CLTV Prediction

Out of the customers selected by data collection, the system calculates the number of customers for the lifetime period under analysis and for a given segment. This information and the data from the CLTV model is then used to calculate the current total profit, the number of customers and lost customers, and the total profit of the following lifetime periods within the specified prediction period.

If you executed the calculation directly, it can take a while for the result to be displayed in table form in the Prediction tab page. The results table lists the different lifetime periods (LTPs) for each prediction period. The prediction period 0 reflects the present, whereas the other prediction periods are used to predict the future developments on the basis of the historic data. The determined prediction is saved to the ODS object CLTV Prediction ( 0CRM_OLVF). You can also display the result with the query CRM CLTV Predictions ( 0CRM_OLVF_Q0001).

The following table shows the result of a CLTV prediction for a sample segment:

Example of CLTV Prediction

Key Date	Prediction Period	Lifetime period	Number of Customers	Lost Customers	Total Profit
01/01/01 (current)	0	1	10		10000
01/01/01	0	2	5		15000
01/01/01	0	3	7		28000
04/01/01 (when LTP = 3 months)	1	2	8	2	8000
04/01/01	1	3	5	0	15000
04/01/01	1	4	6	1	24000

The results allow you to draw the following conclusions:

·        In prediction period 1, I can expect to retain in LTP2 eight of the ten customers currently in LTP1 and to make 8000 USD profit with these customers.

·        I can expect to retain in prediction period 1 all customers currently in LTP2, that is, all five are carried over into LTP3.