System and method for managing routing of customer calls to agents

• US 10,565,529 B1
• Filed: 08/20/2019
• Issued: 02/18/2020
• Est. Priority Date: 08/29/2017
• Status: Active Grant

• Current Assignee: Massachusetts Mutual Life Insurance Company
• Original Assignee: Massachusetts Mutual Life Insurance Company
• Inventors: Merritt, Sears
• Primary Examiner(s): Hong, Harry S
• Application Number:
  US16/546,052
• Time in Patent Office: 182 Days
• Field of Search: None
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/20 Ensemble learning
• CPC Class Codes: G06N 5/01 Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06Q 30/0201 Market modelling; Market analysis; Collecting market data
• CPC Class Codes: G06Q 30/0282 Rating or review of business operators or products
• CPC Class Codes: G06Q 30/0601 Electronic shopping [e-shopping]
• CPC Class Codes: H04M 2203/556 Statistical analysis and interpretation
• CPC Class Codes: H04M 2203/558 Databases
• CPC Class Codes: H04M 3/42059 Making use of the calling party identifier
• CPC Class Codes: H04M 3/42195 Arrangements for calling back a calling subscriber when the wanted subscriber ceases to be busy H04M3/48
• CPC Class Codes: H04M 3/4365 based on information specified by the calling party, e.g. priority or subject
• CPC Class Codes: H04M 3/5183 Call or contact centers with computer-telephony arrangements
• CPC Class Codes: H04M 3/5231 with call back arrangements
• CPC Class Codes: H04M 3/5232 Call distribution algorithms
• CPC Class Codes: H04M 3/5233 Operator skill based call distribution
• CPC Class Codes: H04M 3/5235 Dependent on call type or called number [DNIS]

Abstract

A call management system of a call center retrieves customer demographic data associated with a customer identifier for an inbound caller, i.e., customer. A predictive model including a logistic regression model and tree based model determines a value prediction signal for the identified customer. Based on the value prediction signal determined, the predictive model classifies the identified customer into a first value group or a second value group. The call management system routes a customer classified in the first value group to a first call queue for connection to one of a first pool of call center agents who are authorized to present an offer to purchase a product, and routes a customer classified in the second value group to a second call queue for connection to one of a second pool of call center agents who are not authorized to present the offer to purchase the product.

Claims

1. A processor-based method, comprising:

upon establishing a call session with an identified customer at an inbound call receiving device;

retrieving, by a processor, customer demographic data for the identified customer;

executing, by the processor, a predictive machine-learning model configured to determine a value prediction signal by inputting customer demographic data, wherein the value prediction signal is representative of a likelihood that the identified customer will accept an offer to purchase a product, the predictive machine-learning model classifying the identified customer into a first value group or into a second value group, wherein a modeled value of sale of the product of the first value group is higher than a modeled value of the sale of the product of the second value group;

directing, by the processor, the inbound call receiving device;

to route the identified customer to a first call queue for connection to one of a first pool of call center agents in the event the processor classifies the identified customer into the first value group; and

to route the identified customer to a second call queue for connection to one of a second pool of call center agents in the event the processor classifies the identified customer into the second value group.

2. The processor based method according to claim 1, wherein the first value group comprises customers having a first set of modeled lifetime values of sale of the product, and the second value group comprises customers having a second set of modeled lifetime values of sale of the product, wherein modeled lifetime values in the first set of modeled lifetime values are higher than modeled lifetime values in the second set of modeled lifetime values.

3. The processor based method according to claim 1, wherein the predictive machine-learning model is further configured to determine the value prediction signal by inputting payment data and marketing costs data.

4. The processor based method according to claim 3, wherein the predictive machine-learning model is continually trained using updated customer demographic data, updated payment data, and updated marketing costs data.

5. The processor based method according to claim 3, wherein the predictive machine-learning model is further configured to determine the value prediction signal by inputting lapse data.

6. The processor based method according to claim 1, wherein the first model comprises a logistic regression model.

7. The processor based method according to claim 6, wherein the logistic regression model comprises one of a logistic regression model with l₁ regularization and a logistic regression model with l₂ regularization.

8. The processor based method according to claim 1, wherein the second model comprises a tree based model.

9. The processor based method according to claim 8, wherein the tree based model comprises a random forests ensemble learning method for classification.

10. The processor based method according to claim 1, wherein the first pool of call center agents comprises agents who are authorized to present the offer to purchase the product, and the second pool of call center agents comprises agents who are not authorized to present the offer to purchase the product.

11. A processor based method for managing customer calls within a call center, comprising:

retrieving, by a processor, customer demographic data associated with a customer identifier for an identified customer in a customer call, via a lookup tool executing on the processor to perform real time matching of customer demographic data to the customer identifier for the identified customer in the customer call;

determining, by a predictive model executing on the processor, a value prediction signal representative of a likelihood that the identified customer will accept an offer to purchase a product, wherein the predictive model comprises a first model operating in conjunction with a second model;

classifying, by the predictive model executing on the processor based on the value prediction signal determined by the predictive model, the identified customer into one of a first value group and a second value group, wherein a modeled value of sale of the product of the first value group is higher than a modeled value of the sale of the product of the second value group; and

in the event the classifying step classifies the identified customer into the first value group, routing, by the processor, the identified customer to a first call queue for connection to one of a first pool of call center agents;

in the event the classifying step classifies the identified customer into the second value group, routing, by the processor, the identified customer to a second call queue for connection to one of a second pool of call center agents.

12. The processor based method according to claim 11, wherein the first value group comprises customers having a first set of modeled lifetime values of sale of the product, and the second value group comprises customers having a second set of modeled lifetime values of sale of the product, wherein modeled lifetime values in the first set of modeled lifetime values are higher than modeled lifetime values in the second set of modeled lifetime values.

13. The processor based method according to claim 11, wherein the first model comprises a logistic regression model and the second model comprises a tree based model.

14. The processor based method according to claim 13, wherein the logistic regression model employs one of l₁ regularization and l₂ regularization.

15. The processor based method according to claim 13, wherein the tree based model comprises a random forests ensemble learning method for classification.

16. The processor based method according to claim 13, wherein the predictive machine-learning model is configured to determine the value prediction signal by inputting customer demographic data, payment data, and marketing costs data into the logistic regression model operating in conjunction with the tree based model.

17. The processor based method according to claim 11, wherein the first pool of call center agents comprises agents who are authorized to present the offer to purchase the product, and the second pool of call center agents comprises agents who are not authorized to present the offer to purchase the product.

18. A system for managing customer calls within a call center, comprising:

an inbound telephone call receiving device for receiving a customer call to the call center;

non-transitory machine-readable memory that stores financial information of the call center for historical product sales;

a predictive modeling module that stores a predictive model of customer value, wherein the predictive model comprises a first model operating in conjunction with a second model; and

a processor, configured to execute an inbound queue management module, wherein the processor in communication with the non-transitory machine-readable memory and the predictive models module executes a set of instructions instructing the processor to;

retrieve customer demographic data associated with a customer identifier for an identified customer in the customer call received by the inbound telephone call receiving device;

retrieve from the non-transitory machine readable memory the financial information of the call center for historical product sales;

determine a value prediction signal for the identified customer via analysis by the predictive model of the customer demographic data associated with the customer identifier for the identified customer and the financial information of the call center for historical product sales;

wherein the value prediction signal is representative of a likelihood that the identified customer will accept an offer to purchase a product;

classify the identified customer into one of a first value group and a second value group based on the value prediction signal, wherein a modeled value of sale of the product of the first value group is higher than a modeled value of the sale of the product of the second value group; and

direct the inbound telephone call receiving device;

in the event the inbound queue management module classifies the identified customer into the first value group, to route the identified customer to a first call queue for connection to one of a first pool of call center agents;

in the event the inbound queue management module classifies the identified customer into the second value group, to route the identified customer to a second call queue for connection to one of a second pool of call center agents.

19. The system according to claim 18, wherein the first model is a logistic regression model and the second model is a tree based model.

20. The system according to claim 19, wherein the logistic regression model comprises one of a logistic regression model with l₁ regularization and a logistic regression model with l₂ regularization, and the tree based model is a random forests ensemble learning method for classification.