Next product purchase and lapse predicting tool

1. A processor-based method, comprising:

executing, by a processor, a predictive machine learning model configured to determine, for each customer record of a plurality of customer records including customer profile data stored in a customer database, a set of product purchase ranks by inputting customer purchase history information and the customer profile data into a regression model, the predictive machine learning model outputting a first subset of the plurality of customer records into a target group and a second subset of the plurality of customer records into a non-target group based upon the set of product purchase ranks,

wherein each of the set of product purchase ranks is representative of a likelihood that a respective customer will accept an offer to purchase a respective product from a set of products,

wherein the predictive machine learning model comprises a plurality of component predictive machine learning models, wherein each of the component predictive machine learning models targets likelihood of purchasing one of the respective products from the set of products, and wherein each of the component predictive models determines a respective one of the set of product purchase ranks, and

wherein the predictive machine learning model is continuously trained using updated customer profile data and updated customer purchase history information; and

running, by the processor, the predictive machine learning module to update and display, by a display device in operative communication with the processor, a graphical user interface (“GUI”) including the first subset of the plurality of customer records in the target group, and further including, for each of the first subset of the plurality of customer records, a selected product from the set of products selected by applying a classification criterion to the set of product purchase ranks.

2. The processor-based method of claim 1 , wherein the set of products comprises major products and minor products that are subsets of the major products, and wherein each of the component predictive machine learning models targets one of the major products from the set of products.

3. The processor-based method of claim 1 , further comprising the step of retrieving product data from an internal product database, wherein the GUI displayed by the user interface further includes the retrieved product data for the selected product for each of the first subset of the plurality of customer records.

4. The processor-based method of claim 1 , wherein the customer purchase history information tracks, for each of the plurality of customer records stored in the customer database, an initial purchase of one of the set of products and a date of the initial purchase, and new and cumulative purchases of products during customer-years commencing from anniversaries of the date of initial purchase.

5. The processor-based method of claim 1 , wherein the plurality of customer records stored in the customer database comprise customer records for a book of business of an agent, and wherein in the step of running the predictive machine learning module to update and display the GUI comprises a customer relationship management (“CRM”) dashboard including information on recommended products for the first subset of the plurality of customer records in the target group.

6. The processor-based method of claim 1 , wherein the processor is a server computer, and the GUI is displayed to a user by a client device displaying a CRM dashboard.

7. The processor-based method of claim 1 , wherein the predictive machine learning model is further configured to determine, for each of the plurality of customer records, a lapse rank representative of a likelihood of lapse in payment for one or more previously purchased product and to output a third subset of the plurality of customer records with highest values of the lapse rank, wherein the GUI further includes the third subset of the plurality of customer records with the highest values of the lapse rank.

8. The processor-based method of claim 7 , wherein the predictive machine learning model is further configured to determine the lapse rank for each of the plurality of customer records by inputting information on history of payments into the regression model utilizing gradient boosting, wherein the GUI further comprises the information on history of payments for each of the third subset of the plurality of customer records with the highest values of the lapse rank.

9. The processor-based method of claim 7 , wherein the GUI further includes modeled lifetime values for each of the third subset of the plurality of customer records based upon customer retention assumptions.

10. The processor-based method of claim 1 , wherein the classification criterion for selecting the product from the set of products for each of the first subset of the plurality of customer records in the target group comprises one or more of: size of the respective customer record's highest product purchase rank relative to a threshold; a tier corresponding to the respective customer record's highest product purchase rank, selected from a plurality of tiers based upon a distribution of product purchase scores for the plurality of customer records; or a percentile classification of the respective customer record's highest product purchase rank relative to product purchase scores for the plurality of customer records.

11. The processor-based method of claim 1 , wherein the predictive machine learning model is configured to determine the set of product purchase ranks by inputting the customer purchase history information and the customer profile data into the regression model utilizing gradient boosting.

12. A processor-based method, comprising:

executing, by a processor, a predictive machine learning model configured to determine, for each customer record of a plurality of customer records including customer profile data stored in a customer database, a set of product purchase ranks by inputting customer purchase history information and the customer profile data into a regression model utilizing gradient boosting, the predictive machine learning model outputting a first subset of the plurality of customer records into a target group and a second subset of the plurality of customer records into a non-target group based upon the set of product purchase ranks,

wherein each of the set of product purchase ranks is representative of a likelihood that a respective customer will accept an offer to purchase a respective product from a set of products,

wherein the predictive machine learning model comprises a plurality of component predictive machine learning models, wherein each of the component predictive machine learning models targets likelihood of purchasing one of the respective products from the set of products, and each of the component predictive machine learning models determines a respective one of the set of product purchase ranks,

wherein the customer purchase history information for each of the plurality of customer records includes an initial product purchase from the set of products, and a date of the initial product purchase, and

wherein the predictive machine learning model is continuously trained using updated customer profile data and updated customer purchase history information; and

updating, by the processor, the plurality of customer records in the customer database to indicate whether the respective customer record is included in the target group or is included in the non-target group.

13. The processor-based method of claim 12 , wherein the customer purchase history information for each of the plurality of customer records further includes new and cumulative purchases of products during customer-years commencing from anniversaries of the date of initial purchase.

14. The processor-based method of claim 12 , wherein the updating each of the plurality of customer records in the customer database further includes for each of the first subset of the plurality of customer records, a selected product from the set of products selected by applying a classification criterion to the set of product purchase ranks.

15. A system, comprising:

non-transitory machine-readable memory that stores customer records comprising customer profile data for a plurality of customers and that stores customer purchase history information comprising information on previous purchase by each customer of one or more products from a set of products, and that further stores product data for the set of products;

a predictive machine learning model configured to determine, for each of the plurality of customer records, a set of product purchase ranks by applying a regression model, wherein each of the set of product purchase ranks is representative of a likelihood that a respective customer will accept an offer to purchase a respective product from the set of products, wherein the predictive machine learning model comprises a plurality of component predictive machine learning models, wherein each of the component predictive machine learning models targets likelihood of purchasing one of the respective products from the set of products, wherein the predictive machine learning model is continuously trained using updated customer profile data and updated customer purchase history information; and

a processor in communication with the non-transitory machine-readable memory and the predictive machine learning model configured to execute a set of instructions instructing the processor to:

for each of the plurality of customer records,

determine the set of product purchase ranks by inputting the customer purchase history information and the customer profile data into the predictive machine learning model;

retrieve the product data for a selected product selected by the highest product purchase rank from the set of product purchase ranks determined;

output a first subset of the plurality of customer records into a target group and a second subset of the plurality of customer records into a non-target group based upon the set of product purchase ranks determined; and

run the predictive machine learning model to update a GUI including the first subset of the plurality of customer records in the target group, and further including, for each of the first subset of the plurality of customer records, the selected product from the set of products selected by the highest product rank, and the product data for the selected product.

16. The system according to claim 15 , further comprising a client device, wherein the processor is a server computer and the GUI is displayed to a user by the client device.

17. The system according to claim 16 , further comprising a CRM platform, wherein the GUI is displayed to the user via a dashboard of the CRM platform.

18. The system according to claim 15 , wherein the predictive machine learning model is further configured to determine, for each of the plurality of customer records, a lapse score representative of a likelihood that the respective customer will lapse in payment for the one or more products previously purchased by the customer; and wherein the set of instructions further instructs the processor to classify each of the plurality of customer records into one of a target retention group with highest values of the lapse scores, and a non-target retention group, and to update and display a GUI of the customer records in the target retention group including the highest lapse scores.