The retail credit debt buyer wants to know the optimal action (call/letter intensity, offer discount and send to legal etc) that maximises the net repayments on the loan, within the business constraints set by user.
Zinia will build deep neural networks for customer payment profile. Zinia then combines these models and simulates for the optimal call/letter intensity that maximises ROI. This optimisation will also tell the client when is the best time to carry out each of these actions.
Zinia has transformed the operations at this retail credit debt buyer. Previously these actions were selected using either credit/internal scores or rules.
First, Zinia automatically cleans and standardizes the data fields and also selects the data variables for the AI model. In this case, Zinia selected the following types of data variables (about 8 final model variables).
|Product features||Other Variables||Loan performance|
|Loan amount||Response to Actions||Paid in full/instalment/No Payment|
|Loan Origination Date||Debt Type|
|Default Date||Internal credit score|
|Other Debt Internally||Payment history|
Zinia will build deep neural networks for customers with different payment profiles. Zinia then combines these models and simulates for the optimal call/letter intensity that maximises ROI. This optimisation also tells the client when is the best time to carry out each of these actions.
Zinia produces daily operational strategy (i.e. which customers to call and letter and when). This leads to an increase of 21% in total net repayments compared to the client’s previous operational strategy (which was based on propensity scoring and rules).