Method for evaluating customer valve to guide loyalty and retention programs

• US 7,340,408 B1
• Filed: 06/13/2000
• Issued: 03/04/2008
• Est. Priority Date: 06/13/2000
• Status: Expired due to Fees

• Current Assignee: Verizon Laboratories Incorporated (Verizon Communications Inc.)
• Original Assignee: Verizon Laboratories Incorporated (Verizon Communications Inc.)
• Inventors: Drew, James Howard, Mani, Denkanikota R., Betz, III, Andrew Louis, Datta-Read, Piew
• Primary Examiner(s): ALLEN, AKIBA KANELLE
• Application Number:
  US09/593,118
• Time in Patent Office: 2,821 Days
• Field of Search: 705/1, 705/7, 705/10, 705/11
• US Class Current: 705/7.33
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0204 Market segmentation

Abstract

A method and apparatus for training a neural network to compute hazard functions for customers and analyzing hazard functions, both for an individual customer, and for set of customers to focus marketing techniques. The hazard function represents the likelihood of churn for a particular customer. The gain in lifetime value is also calculated for each customer which incorporates the present value of the customer with the future value of the customer if a new contract is entered. The overall shape of the hazard function, combined with the gain in lifetime value, specifies what marketing techniques are to be applied together with what additional incentives are to be offered to the customer in order prevent churn.

Claims

1. A method for evaluating customer value to guide loyalty and retention programs comprising:

generating, by a processing system, a hazard function model based on attributes relating to a plurality of current customer accounts;

generating, by the processing system, a hazard function for a first existing customer, to determine probability of churn, based on the hazard function model and account data associated with the first existing customer and corresponding to the attributes;

calculating a gain in lifetime value for the first existing customer based on a change in the hazard function resulting from a retention effort; and

determining a focus for retention-based interactions with the first existing customer based on at least the hazard function, wherein determining the focus comprises clustering the hazard function for the first existing customer and hazard functions for a plurality of other existing customers so that the hazard functions are grouped together according to shape, each group representative of a customer set.

2. The method of claim 1, wherein calculating the gain in lifetime value includes:

calculating a lifetime value based on original contract terms and revenue associated with the first existing customer; and

calculating the gain in lifetime value based on a change in the hazard function resulting from a new contract period.

3. The method of claim 1, wherein determining the focus for retention-based interactions with the first existing customer includes:

analyzing the shape of the hazard function generated for the first existing customer; and

specifying a set of marketing techniques based on the shape of the hazard function.

4. The method of claim 1, wherein determining the focus for retention-based interactions with the first existing customer includes:

specifying a set of incentives to offer the first existing customer based on the gain in lifetime value.

5. The method of claim 3, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, there is no effect on churn of a contract expiration.

6. The method of claim 5, wherein specifying the set of marketing techniques includes:

taking no further steps to deter churn.

7. The method of claim 3, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, that there is a small increase in probability of churn at contract expiration, with an elevated post-expiration churn.

8. The method of claim 7, wherein specifying the set of marketing techniques includes:

having a moderate pre-expiration effort where new contracts or continued contracts are the goal.

9. The method of claim 3, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, that there is a large spike indicating high probability of churn at contract expiration and low probability of churn thereafter.

10. The method of claim 9, wherein specifying the set of marketing techniques includes:

concentrating effort on pre-expiration of contract where a contract renewal may not be required.

11. The method of claim 3, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, that there is a large increase in probability of churn at expiration with high and increasing post-expiration probability of churn.

12. The method of claim 11, wherein specifying the set of marketing techniques includes:

specifying a high intensity pre-expiration effort with continued competitive offers to maintain the first existing customer.

13. The method of claim 4, wherein specifying the incentives includes:

determining that a value of the set of incentives offered to the first existing customer does not exceed the gain in lifetime value.

14. The method of claim 1, wherein determining the focus for retention-based interactions with the first existing customer includes:

determining, based on the overall shape of the clustered hazard functions, a focus for retention-based interactions for each customer set.

62. The method of claim 1, wherein calculating the gain in lifetime value comprises calculating expected revenue multiplied by an increase in remaining lifetime resulting from the retention effort.

15. An apparatus for evaluating customer value to guide loyalty and retention programs comprising:

a computer, comprising;

a first calculating component for generating a hazard function model based on attributes relating to a plurality of current customer accounts;

a generating component for generating a hazard function for a first existing customer, to determine probability of churn, based on the hazard function model and account data associated with the first existing customer and corresponding to the attributes;

a second calculating component for calculating a gain in lifetime value for the first existing customer based on a change in the hazard function resulting from a retention effort;

a determining component for determining a focus for retention-based interactions with the first existing customer based on at least the hazard function; and

a clustering component for clustering the hazard function for the first existing customer and hazard functions for a plurality of other existing customers so that the hazard functions are grouped together according to shape, each group representative of a customer set; and

a display device displaying at least one of the hazard function, the gain in lifetime value for the first existing customer, and the focus for retention-based interactions with the first existing customer.

16. The apparatus of claim 15, wherein the second calculating component for calculating the gain in lifetime value

calculates a lifetime value based on original contract terms and revenue associated with the first existing customer;

and

calculates the gain in lifetime value based on a change in the hazard function resulting from a new contract period.

17. The apparatus of claim 15, wherein the determining component includes:

an analyzing component for analyzing the shape of the hazard function generated for the first existing customer; and

a specifying component for specifying a set of marketing techniques based on the shape of the hazard function.

18. The apparatus of claim 15, wherein the determining component includes:

a specifying component for specifying a set of incentives to offer the first existing customer based on the gain in lifetime value.

19. The apparatus of claim 17, wherein the specifying component for specifying the set of marketing techniques based on the shape includes:

a determining component for determining, based on the shape of the hazard function, there is no effect on churn of a contract expiration.

20. The apparatus of claim 19, wherein the specifying component specifies taking no further steps to deter churn.

21. The apparatus of claim 17, wherein the specifying component for specifying the set of marketing techniques based on the shape includes:

a determining component for determining, based on the shape of the hazard function, that there is a small increase in probability of churn at contract expiration, with an elevated post-expiration churn.

22. The apparatus of claim 21, wherein the specifying component specifies a moderate pre-expiration effort where new contracts or continued contracts are the goal.

23. The apparatus of claim 17, wherein the specifying component for specifying the set of marketing techniques based on the shape includes:

a determining component for determining, based on the shape of the hazard function, that there is a large spike indicating high probability of churn at contract expiration and low probability of churn thereafter.

24. The apparatus of claim 23, wherein the specifying component specifies concentrating effort on pre-expiration of contract where a contract renewal may not be required.

25. The apparatus of claim 17, wherein the specifying component for specifying the set of marketing techniques based on the shape includes:

a determining component for determining, based on the shape of the hazard function, that there is a large increase in probability of churn at expiration with high and increasing post-expiration probability of churn.

26. The apparatus of claim 25, wherein the specifying component specifies a high intensity pre-expiration effort with continued competitive offers to maintain the first existing customer.

27. The apparatus of claim 18, wherein the specifying component includes a determining component for determining that a value of the set of incentives does not exceed the gain in lifetime value.

28. The apparatus of claim 15, wherein the determining component determines, based on the overall shape of the clustered hazard functions, a focus for retention-based interactions for each customer set.

63. The method of claim 15, wherein the second calculating component calculates the gain in lifetime value by calculating expected revenue multiplied by an increase in remaining lifetime resulting from the retention effort.

29. A computer-readable medium including instructions, executable by a processor, for performing a method for evaluating customer value to guide loyalty and retention programs, the method comprising:

generating a hazard function model based on attributes relating to a plurality of current customer accounts;

generating a hazard function for a first existing customer, to determine probability of churn, based on the hazard function model and account data associated with the first existing customer and corresponding to the attributes;

calculating a gain in lifetime value for the first existing customer based on a change in the hazard function; and

determining a focus for retention-based actions based on at least the hazard function, wherein determining the focus comprises clustering the hazard function for the first existing customer and hazard functions for a plurality of other existing customers so that the hazard functions are grouped together according to shape, each group representative of a customer set.

30. The computer-readable medium of claim 29, wherein

calculating the gain in lifetime value includes:

calculating a lifetime value based on original contract terms and revenue associated with the first existing customer; and

calculating the gain in lifetime value based on a change in the hazard function resulting from a new contract period.

31. The computer-readable medium of claim 29, wherein determining the focus for retention-based actions includes:

analyzing the shape of the hazard function generated for the first existing customer; and

specifying a set of marketing techniques based on the shape of the hazard function.

32. The computer-readable medium of claim 29, wherein determining the focus for retention-based actions includes:

specifying a set of incentives offered to the first existing customer based on the gain in lifetime value.

33. The computer-readable medium of claim 31, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, there is no effect on churn of a contract expiration.

34. The computer-readable medium of claim 33, wherein specifying the set of marketing techniques includes:

taking no further steps to deter churn.

35. The computer-readable medium of claim 31, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, that there is a small increase in probability of churn at contract expiration, with an elevated post-expiration churn.

36. The computer-readable medium of claim 35, wherein specifying the set of marketing techniques includes:

having a moderate pre-expiration effort where new contracts or continued contracts are the goal.

37. The computer-readable medium of claim 31, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, that there is a large spike indicating high probability of churn at contract expiration and low probability of churn thereafter.

38. The computer-readable medium of claim 37, wherein specifying the set of marketing techniques includes:

concentrating effort on pre-expiration of contract where a contract renewal may not be required.

39. The computer-readable medium of claim 31, wherein specifying the set of marketing techniques based on the shape includes:

determining, based on the shape of the hazard function, that there is a large increase in probability of churn at expiration with high and increasing post-expiration probability of churn.

40. The computer-readable medium of claim 39, wherein specifying the set of marketing techniques includes:

specifying a high intensity pre-expiration effort with continued competitive offers to maintain the first existing customer.

41. The computer-readable medium of claim 32, wherein specifying the incentives includes:

determining that a value of the set of incentives does not exceed the gain in lifetime value.

42. The computer-readable medium of claim 29, wherein determining the focus for retention-based actions includes:

determining, based on the overall shape of the clustered hazard functions, a focus for retention-based actions for each customer set.

64. The computer-readable medium of claim 29, wherein calculating the gain in lifetime value comprises calculating expected revenue multiplied by an increase in remaining lifetime resulting from a retention effort.

43. A system for evaluating customer value to guide loyalty and retention programs comprising:

means for generating a hazard function model based on attributes relating to a plurality of current customer accounts;

means for generating a hazard function for a first existing customer, to determine probability of churn, based on the hazard function model and account data associated with the first existing customer and corresponding to the attributes;

means for calculating a gain in lifetime value for the first existing customer based on a change in the hazard function;

means for determining a focus for retention-based actions based on at least the hazard function;

means for clustering the hazard function for the first existing customer and hazard functions for a plurality of other existing customers so that the hazard functions are grouped together according to shape, each group representative of a customer set; and

means for displaying at least one of the hazard function, the gain in lifetime value for the first existing customer, and the focus for retention-based actions.

44. The system of claim 43, wherein the means for calculating the gain in lifetime value includes:

means for calculating a lifetime value based on original contract terms and revenue associated with the first existing customer; and

means for calculating the gain in lifetime value based on a change in the hazard function resulting from a new contract period.

45. The system of claim 43, wherein the means for determining includes:

means for analyzing the shape of the hazard function generated for the first existing customer; and

means for specifying a set of marketing techniques based on the shape of the hazard function.

46. The system of claim 43, wherein the means for determining includes:

means for specifying a set of incentives to offer the first existing customer based on the gain in lifetime value.

47. The system of claim 45, wherein the means for specifying the set of marketing techniques based on the shape includes:

means for determining, based on the shape of the hazard function, that there is no effect on churn of a contract expiration.

48. The system of claim 47, wherein the means for specifying the set of marketing techniques specifies taking no further steps to deter churn.

49. The system of claim 45, wherein the means for specifying the set of marketing techniques based on the shape includes:

means for determining, based on the shape of the hazard function, that there is a small increase in probability of churn at contract expiration, with an elevated post-expiration churn.

50. The system of claim 49, wherein the means for specifying the set of marketing techniques specifies a moderate pre-expiration effort where new contracts or continued contracts are the goal.

51. The system of claim 45, wherein the means for specifying the set of marketing techniques based on the shape includes:

means for determining, based on the shape of the hazard function, that there is a large spike indicating high probability of churn at contract expiration and low probability of churn thereafter.

52. The system of claim 51, wherein the means for specifying the set of marketing techniques specifies concentrating effort on pre-expiration of contract where a contract renewal may not be required.

53. The system of claim 45, wherein the means for specifying the set of marketing techniques based on the shape includes:

means for determining, based on the shape of the hazard function, that there is a large increase in probability of churn at expiration with high and increasing post-expiration probability of churn.

54. The system of claim 53, wherein the means for specifying the set of marketing techniques specifies a high intensity pre-expiration effort with continued competitive offers to maintain the first existing customer.

55. The system of claim 46, wherein the means for specifying the incentives includes:

means for determining that a value of the set of incentives does not exceed the gain in lifetime value.

56. The system of claim 43, wherein the means for determining the focus for retention-based actions includes:

means for determining, based on the overall shape of the clustered hazard functions, a focus for retention-based actions for each customer set.

65. The system of claim 43, wherein the means for calculating a gain in lifetime value calculates the gain in lifetime value based on a change in the hazard function by calculating expected revenue multiplied by an increase in remaining lifetime resulting from a retention effort.

57. A method for evaluating customer value to guide loyalty and retention programs comprising:

generating, by a multilayer feed-forward neural network, a hazard function for a first existing customer, to determine probability of churn, based on account data associated with the first existing customer and corresponding to a set of attributes;

calculating, for the first existing customer, a gain in lifetime value based on a change in the hazard function resulting from a retention effort; and

determining a focus for a retention-based program based on at least the hazard function, wherein determining the focus comprises clustering the hazard function for the first existing customer and hazard functions for a plurality of other existing customers so that the hazard functions are grouped together according to shape, each group representative of a customer set.

58. The method of claim 57 further comprising:

implementing the programs based on the determined focus.

66. The method of claim 57, further comprising:

training the neural network to generate a hazard function model based on account data associated with a plurality of current customer accounts and corresponding to the set of attributes; and

wherein generating a hazard function includes generating a hazard function for an existing customer, to determine probability of churn, based on the hazard function model and the account data associated with the customer and corresponding to a set of attributes.

67. The method of claim 66, wherein training the neural network comprises loading an input layer of the neural network with values representing the set of attributes for the plurality of current customer accounts.

59. A method for evaluating customer value to guide loyalty and retention programs comprising:

generating, by a processing system and for each of a plurality of customers, a hazard function to determine a probability of churn for each customer, the hazard function based on attributes relating to customer account information;

identifying a temporal-based retention effort based on the hazard function for each of the plurality of customers;

calculating, for each of the plurality of customers, an expected gain in value from the identified retention effort; and

determining a focus for customer interaction based on the expected gain in value, wherein determining the focus comprises clustering the hazard function for each of the plurality of customers so that the hazard functions are grouped together according to shape, each group representative of a customer set.

60. The method of claim 59, wherein generating the hazard function comprises:

generating the hazard function, based on a reference hazard function model, for each of the plurality of customers.

61. The method of claim 59, wherein the temporal-based retention effort comprises retention actions directed to each customer during a first time period and retention actions directed to each customer during a second time period occurring after the first time period.