Predictive call routing

• US 8,488,774 B2
• Filed: 07/23/2007
• Issued: 07/16/2013
• Est. Priority Date: 07/24/2006
• Status: Active Grant

• Current Assignee: FMR LLC
• Original Assignee: FMR LLC
• Inventors: Mahalaha, Anil, Tonelli, Daniel L., Berk, Barry
• Primary Examiner(s): Tsang, Fan
• Assistant Examiner(s): Intavong, Jirapon
• Application Number:
  US11/781,906
• Publication Number:
  US 20080095355A1
• Time in Patent Office: 2,185 Days
• Field of Search: 379/265.01, 379/265.02, 379/265.06, 379/265.09
• US Class Current: 379/265.13
• CPC Class Codes: H04M 2201/12 Counting circuits
• CPC Class Codes: H04M 2201/36 Memories
• CPC Class Codes: H04M 3/5233 Operator skill based call distribution

Abstract

A predictive call routing system that includes a real-time decision engine to receive information about a customer and identify a skill that is useful for providing service to the caller. The decision engine identifies the skill by generating scores for a plurality of statistical models using the statistical models and parameters associated with the caller, each statistical model representing a correlation between a subset of parameters and an action that may be performed or requested to be performed by the caller, the score for each statistical model being generated using the statistical model and the subset of parameters associated with the statistical model, and identifies a skill based on the scores. The system includes a call router to route a call from the customer to a representative who has the skill.

Claims

1. A method comprising:

at a call routing system, receiving a telephone call from a caller, the caller being associated with a plurality of parameters;

predicting an action prior to caller input about the action, including generating scores for a plurality of statistical models, each statistical model representing a correlation between a subset of the plurality of parameters and an action, the score for each statistical model being generated using the statistical model and the subset of the plurality of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the caller will perform an action associated with the statistical model or request the action to be performed;

selecting a representative based on the scores; and

routing the call to the selected representative.

2. The method of claim 1 wherein selecting a representative comprises selecting a statistical model based on the scores and predetermined rules.

3. The method of claim 2 wherein selecting a representative comprises mapping the selected statistical model to a skill, and selecting the representative based on the skill.

4. The method of claim 3 wherein the skill is useful in performing the action.

5. The method of claim 3 wherein mapping the selected statistical model to a skill comprises using predetermined rules to map the selected statistical model to a skill.

6. The method of claim 3 wherein mapping the selected statistical model to a skill comprises using a second layer of statistical analyses, the second layer of statistical analyses comprising

generating a second set of scores for a plurality of second statistical models, each second statistical model being associated with a skill, and

mapping the selected statistical model to the skill based on the second set of scores.

7. The method of claim 1 wherein the statistical models are derived from information collected on people who have previously called using telephone numbers that are handled by the call routing system.

8. The method of claim 7 wherein the statistical model associated with a particular action is derived by finding previous calls that are associated with the particular action, and analyzing parameters of the callers to determine which subset of parameters correlate to the particular action.

9. The method of claim 1 wherein generating scores for the plurality of statistical models comprises generating scores for some of the statistical models, in which the statistical models comprise nested layers of statistical models, each model being linked to one or more models in the next layer.

10. The method of claim 1 wherein the action comprises at least one of requesting information, opening an account, updating account information, and performing a transaction with respect to an account.

11. The method of claim 1, further comprising receiving a feedback indicating whether it was proper to route the call to the selected representative.

12. The method of claim 11, further comprising updating the statistical models based on the feedback.

13. The method of claim 1, further comprising generating and broadcasting a personalized message to the caller while placing the caller on hold and waiting for a representative, wherein the personalized message is generated based on the scores.

14. The method of claim 1 wherein the parameters associated with the caller comprises at least one of age of the caller, gender of the caller, marital status of the caller, assets in the caller'"'"'s account, number of previous calls from the caller within a predetermined number of days, number of days since a promotional material was sent to the caller, whether the caller has visited particular web pages, and language preferences of the caller.

15. The method of claim 1, further comprising searching a database having information about representatives, skills of the representatives, and proficiency of the representatives with respect to the skills.

16. The method of claim 1 wherein the parameters comprise whether the caller has a financial plan that is the subject of a news article published within a predetermined number of days.

17. The method of claim 1 wherein the selection of a representative is intended to maintain a freshness of training for the representative by distributing calls among representatives.

18. A method comprising:

collecting information about past callers who have called one or more specified telephone numbers and performed an action or requested the action to be performed;

analyzing the information to determine which parameters of the callers correlate to the action;

establishing a statistical model based on the parameters that correlate to the action, wherein a score that is generated by using the statistical model and parameters of a caller provides information about a probability that the caller will perform the action or request the action to be performed;

at a call routing system, receiving a telephone call from a caller of the telephone call, the caller being associated with a plurality of parameters; and

predicting, using the statistical model and other statistical models, an action prior to caller input about the action.

19. A method comprising:

receiving a call from a customer, the customer being associated with a plurality of parameters;

predicting an action prior to customer input about the action, including generating scores for a plurality of statistical models, each statistical model representing a correlation between a subset of the plurality of parameters and an action, the score for each statistical model being generated using the statistical model and the subset of the plurality of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the customer will perform an action associated with the statistical model or request the action to be performed;

providing, based on the scores, a subset of most likely service options to the customer; and

prompting the customer to select an option from the subset of service options.

20. A method comprising:

receiving a telephone call from a customer, the customer being associated with a plurality of parameters;

predicting an action prior to customer input about the action, including generating scores for a plurality of statistical models, each statistical model representing a correlation between a subset of the plurality of parameters and an action, the score for each statistical model being generated using the statistical model and the subset of the plurality of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the customer will perform an action associated with the statistical model or request the action to be performed;

generating an e-mail response based on the scores; and

generating a voice message to tell the customer that an e-mail having information useful to the customer will be sent to the customer.

21. A method comprising:

receiving a telephone call from a customer, the customer being associated with a plurality of parameters;

predicting an action prior to customer input about the action, including generating scores for a plurality of statistical models, each statistical model representing a correlation between a subset of the plurality of parameters and an action, the score for each statistical model being generated using the statistical model and the subset of the plurality of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the customer will perform an action associated with the statistical model or request the action to be performed;

determining a criterion based on the scores; and

determining that a representative meeting the criterion is not currently available.

22. The method of claim 21, comprising generating a voice message to tell the customer that a representative will call back at a later time.

23. The method of claim 22, comprising prompting the customer to provide a time and date for the representative to call back.

24. A method comprising:

receiving a call from a customer;

broadcasting options of a main menu to the customer;

upon receiving a customer selection indicating that the customer wishes to speak to a representative, predicting an action prior to caller input about the action, providing a tutorial to the customer based on the prediction, the tutorial selected based on scores that are generated using a plurality of statistical models and parameters associated with the customer, each statistical model representing a correlation between a subset of parameters and an action, the score for each statistical model being generated using the statistical model and the subset of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the caller will perform an action associated with the statistical model or request the action to be performed;

transferring the call to a representative; and

notifying the representative that the customer has been given the tutorial.

25. A method comprising:

determining a skill of a representative for servicing a customer using two layers statistical modeling, wherein the first layer of statistical modeling provides estimations of actions that the customer is likely to perform or request to be performed before the customer provides input about the actions, the first layer of statistical modeling involving generating scores for a plurality of statistical models in the first layer, in which the score for each statistical model in the first layer provides information about a probability that the caller will perform an action associated with the statistical model or request the action to be performed, and the second layer of statistical modeling provides estimations of skills that are likely useful in performing the actions.

26. The method of claim 25, further comprising determining a proficiency level of the skill also using the two layers statistical modeling.

27. A method comprising:

choosing one option out of a plurality of options based on at least two layers of statistical modeling that comprises

generating a first set of scores for a first set of statistical models using the first set of statistical models and a first set of parameters, the score for each of the first set of statistical models being generated using the statistical model and a subset of the first set of parameters associated with the statistical model, the score for each statistical model providing information about a probability that a user will perform an action associated with the statistical model or request the action to be performed;

selecting a score from among the first set of scores prior to receiving input from the user about an action associated with the statistical model corresponding to the score;

generating a second set of scores for a second set of statistical models using the second set of statistical models and a second set of parameters, the second set of parameters including the selected score from the first set of scores, the score for each of the second set of statistical models being generated using the statistical model and a subset of second set of parameters associated with the statistical model; and

choosing the option based on the second set of scores.

28. The method of claim 27 wherein choosing one option from among the plurality of options comprises selecting a representative from among a plurality of representatives, the selected representative having the skill and proficiency level for servicing a caller.

29. A system comprising:

a real-time decision engine to receive information about a caller and identify a skill that is useful for providing service to the caller, the caller being associated with a plurality of parameters, the decision engine identifying the skill by

predicting an action prior to caller input about the action, including generating scores for a plurality of statistical models, each statistical model representing a correlation between a subset of the plurality of parameters and an action that may be performed or requested to be performed by the caller, the score for each statistical model being generated using the statistical model and the subset of the plurality of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the caller will perform an action associated with the statistical model or request the action to be performed, and

identifying a skill based on the scores;

a storage to store the statistical models; and

a call router to route a call from the customer to a representative who has the skill.

30. The system of claim 29, further comprising a storage to store a database having information about customers.

31. The system of claim 29, further comprising a storage to store a database having information about representatives and skills associated with the representatives.

32. The system of claim 29 wherein the decision engine is configured to establish a statistical model associated with a particular action by finding previous calls that are associated with the particular action, and finding parameters of callers to determine which subset of parameters correlate to the particular action.

33. The system of claim 29 wherein the plurality of statistical models comprise nested layers of statistical models, each model being linked to one or more models in the next layer.

34. The system of claim 29 wherein identifying the skill based on the scores comprises using a second layer of statistical modeling to determine the skill based on the scores.

35. The system of claim 34 wherein the second layer of statistical modeling comprises generating scores for a second set of plurality of statistical models, each of the second set of statistical models representing a correlation between a statistical model in the first layer, a subset of the plurality of parameters, and a skill that may be useful for providing service to the caller that performs or requests the action associated with the statistical model in the first layer.

37. The system of claim 29 in which the real-time decision engine uses feedback indicating whether it was proper to route the call to the representative to update the statistical models.

36. A call routing system comprising:

a real-time decision engine to receive information about a caller and identify a group of service options that are useful to the caller, the caller being associated with a plurality of parameters, the decision engine identifying the service options by

predicting an action prior to caller input about the action, including generating scores for a plurality of statistical models, each statistical model representing a correlation between a subset of the plurality of parameters and an action that may be performed or requested to be performed by the caller, the score for each statistical model being generated using the statistical model and the subset of the plurality of parameters associated with the statistical model, the score for each statistical model providing information about a probability that the caller will perform an action associated with the statistical model or request the action to be performed, and

identifying a subset of most likely service options based on the scores;

a storage to store the statistical models; and

an interactive voice response system to provide the subset of most likely service options to the caller and prompt the caller to select one of the service options.