Managing customer loss using customer groups

• US 7,813,952 B2
• Filed: 06/03/2003
• Issued: 10/12/2010
• Est. Priority Date: 06/04/2002
• Status: Active Grant

• Current Assignee: SAP SE
• Original Assignee: SAP AG (SAP SE)
• Inventors: Eskandari, Ramine
• Primary Examiner(s): Jeanty; Romain
• Assistant Examiner(s): MANSFIELD, THOMAS L
• Application Number:
  US10/452,911
• Publication Number:
  US 20040073520A1
• Time in Patent Office: 2,688 Days
• Field of Search: 705/10
• US Class Current: 705/7.31
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/0283 Price estimation or determination

Abstract

Techniques are provided for managing customer loss. Customers are first grouped using a predetermined category definition and then customers in one group are segmented based on common customer characteristics. The techniques may be used to categorize customers based on a likelihood of being lost and segmenting customers with a high likelihood of being lost into smaller, more homogenous groups of customers based on shared customer characteristics.

Claims

1. A computer-implemented method for managing customer loss, the method comprising:

accessing, from electronic storage, customer information having multiple customer records, each customer record including multiple attribute values;

determining, using at least one processor, for each customer for which a record was accessed a churn likelihood representing the probability that the customer will be lost within a predetermined period of time;

computing, for each customer for which a record was accessed and using at least one processor, a composite-customer-value index for the customer as a sum of (1) a product-cost profitability measure for the customer determined by subtracting product costs from the customer'"'"'s net sales, (2) a sales-cost profitability measure for the customer determined by subtracting, from the product-cost profitability measure, additional direct and indirect sales costs associated with the customer, and (3) a result of dividing the sales-cost profitability measure for the customer by the product-cost profitability measure for the customer, the composite-customer-value index representing the value of the customer to a business enterprise;

for each customer for which a record was accessed, using at least one processor to normalize the composite-customer-value index for the customer and apply a first statistical coefficient to the composite-customer-value index for the customer to calculate a processed composite-customer-value index for the customer;

for each customer for which a record was accessed, using at least one processor to apply a second statistical coefficient to the churn likelihood for the customer to calculate a processed churn likelihood index for the customer, the second statistical coefficient being different than the first statistical coefficient;

for each customer for which a record was accessed, computing, using at least one processor, a value-churn measure for the customer the processed churn likelihood index for the customer and the processed composite-customer-value index for the customer;

accessing, for each of at least three likelihood-to-churn categories and using at least one processor, a predetermined threshold that identifies a range of value-churn measures to be used for determining customers to be associated with a particular likelihood-to-churn category from among the at least three likelihood-to-churn categories;

analyzing, using at least one processor, the determined value-churn measure for each customer with respect to the accessed predetermined thresholds;

based on the analysis, associating, using at least one processor, each customer with one of the at least three likelihood-to-churn categories, the association being based on the determined value-churn measure of a customer falling within the range of value-churn measures corresponding to the associated likelihood-to-churn category;

receiving, using at least one processor, user input selecting, from among the at least three likelihood-to-churn categories, a likelihood-to-churn category to be used to cluster customers that are associated with the selected likelihood-to-churn category;

based on receiving user input selecting a likelihood-to-churn category to be used to cluster customers, clustering, using at least one processor, customers associated with the selected likelihood-to-churn category into multiple groups of customers having a shared characteristic that is different than likelihood-to-churn and value to the business enterprise; and

defining, for each of the multiple groups of customers and using at least one processor, a promotional campaign that is targeted to customers having a value-churn measure falling within the range of value-churn measures corresponding to the selected likelihood-to-churn category and that is targeted to the shared characteristic of the customers clustered into the corresponding group.

2. The method of claim 1 further comprising taking action for the purpose of improving the likelihood that a customer will be retained, the action taken being based on the association of the customer with a customer group.

3. The method of claim 1 further comprising:

identifying, for each customer represented by a customer record, an importance value that represents the value of the customer to a business enterprise; and

identifying customer records that have both a high churn likelihood and a high importance value, wherein associating each customer in one of the likelihood-to-churn categories with one of several customer groups comprises associating each customer that has both a high churn likelihood and a high importance value into one of several customer groups.

4. The method of claim 3 wherein the importance value comprises an importance value having at least two importance indicators.

5. The method of claim 3 wherein the importance value comprises a profitability value that represents the contribution of the customer to the business enterprise.

6. The method of claim 5 wherein the profitability value comprises a profitability value having 1) a product-cost value that represents a net sales-cost value arrived at by subtracting a sales deductions value from a gross sales value and 2) a sales-cost value arrived at by subtracting an additional cost value associated with selling to the customer from the product-cost value.

7. The method of claim 5 wherein the sales-cost value comprises a direct sales-cost value arrived at by subtracting a direct sales-cost value associated with selling to the customer from the product-cost value.

8. The method of claim 5 wherein the sales-cost value comprises an indirect sales-cost value arrived at by subtracting an indirect sales-cost value associated with selling to the customer from the product-cost value.

9. The method of claim 5 further comprising:

applying a first statistical weight to the product-cost value; and

applying a second statistical weight to the sales-cost value, wherein the profitability value comprises a profitability value based on the application of a first statistical weight to the product-cost value and the application of a second statistical weight to sales-cost value.

10. The method of claim 9 wherein the first statistical weight is the same as the second statistical weight.

11. The method of claim 9 wherein the first statistical weight is different from the second statistical weight.

12. The method of claim 9 wherein the first statistical weight and the second statistical weight are user-configurable.

13. The method of claim 1 further comprising:

generating a data model that predicts the likelihood that each customer will be lost within a predetermined period of time; and

applying the data model to the accessed customer information to determine the churn likelihood for each customer for which a record was accessed.

23. The method of claim 1 wherein:

accessing, for each of at least three likelihood-to-churn categories, a predetermined threshold that identifies a range of value-churn measures to be used for determining customers to be associated with a particular likelihood-to-churn category from among the at least three likelihood-to-churn categories comprises accessing a predetermined threshold for each of at least three likelihood-to-churn categories including a relatively high value-churn category that identifies a range of value-churn measures indicating a relatively high combination of likelihood-to-churn and value to the business enterprise, a relatively medium value-churn category that identifies a range of value-churn measures indicating a relatively medium combination of likelihood-to-churn and value to the business enterprise, and a relatively low value-churn category that identifies a range of value-churn measures indicating a relatively low combination of likelihood-to-churn and value to the business enterprise;

analyzing the determined value-churn measure for each customer with respect to the accessed predetermined thresholds comprises analyzing the determined value-churn measure for each customer with respect to the accessed predetermined thresholds for the relatively high value-churn category, the relatively medium value-churn category, and the relatively low value-churn category;

associating each customer with one of the at least three likelihood-to-churn categories comprises associating each customer with one of the at least three likelihood-to-churn categories including the relatively high value-churn category, the relatively medium value-churn category, and the relatively low value-churn category;

receiving user input selecting, from among the at least three likelihood-to-churn categories, a likelihood-to-churn category to be used to cluster customers that are associated with the selected likelihood-to-churn category comprises receiving user input selecting a likelihood-to-churn category from among the relatively high value-churn category, the relatively medium value-churn category, and the relatively low value-churn category, the selection including less than all of the relatively high value-churn category, the relatively medium value-churn category, and the relatively low value-churn category;

clustering, using at least one processor, customers associated with the selected likelihood-to-churn category into multiple groups of customers having a shared characteristic that is different than likelihood-to-churn and value to the business enterprise comprises using multivariate analysis of demographic characteristics of the customers associated with the selected likelihood-to-churn category to cluster each of the customers associated with the selected likelihood-to-churn category into only one of multiple homogenous groups of customers having shared demographic characteristics that are different than likelihood-to-churn and value to the business enterprise; and

defining, for each of the multiple groups of customers, a promotional campaign that is targeted to customers having a value-churn measure falling within the range of value-churn measures corresponding to the selected likelihood-to-churn category and that is targeted to the shared characteristic of the customers clustered into the corresponding group comprises defining, for each of the multiple homogenous groups of customers, a different promotional campaign that is targeted to customers having a value-churn measure falling within the range of value-churn measures corresponding to the selected likelihood-to-churn category and that is focused on the shared demographic characteristics of the customers clustered into the corresponding homogenous group.

24. The method of claim 1, wherein computing, for each customer for which a record was accessed and using at least one processor, a composite-customer-value index for the customer comprises:

determining a number of referrals made by a customer; and

accounting for the determined number of referrals in computing the composite-customer-value index for the customer.

25. The method of claim 1, wherein computing, for each customer for which a record was accessed and using at least one processor, a composite-customer-value index for the customer comprises:

determining a length of time a customer has had a relationship with the business enterprise; and

accounting for the determined length of time the customer has had a relationship with the business enterprise in computing the composite-customer-value index for the customer.

26. The method of claim 1, wherein computing, for each customer for which a record was accessed and using at least one processor, a composite-customer-value index for the customer comprises:

determining a product-cost profitability measure for the customer by;

determining gross sales for the customer, determining deductions for the customer, determining the customer'"'"'s net sales by subtracting the determined deductions for the customer from the determined gross sales for the customer, and subtracting the product costs from the customer'"'"'s net sales;

determining a sales-cost profitability measure for the customer by;

determining direct sales costs associated with the customer, the direct sales costs relating to sales costs attributed directly to making a particular sale to the customer and including at least marketing costs and promotional costs, customer-related order costs, and customer-related shipment costs, determining indirect sales costs associated with the customer, the indirect sales costs relating to sales costs not attributed directly to making a particular sale to the customer and including at least costs for customer visits, customer support, and customer care, subtracting the determined direct sales costs and the determined indirect sales costs from the determined product-cost profitability measure;

dividing the determined sales-cost profitability measure for the customer by the determined product-cost profitability measure for the customer to produce the result of dividing the sales-cost profitability measure for the customer by the product-cost profitability measure for the customer; and

computing the composite-customer-value index for the customer by summing the determined product-cost profitability measure for the customer, the determined sales-cost profitability measure for the customer, and the result of dividing the sales-cost profitability measure for the customer by the product-cost profitability measure for the customer.

27. The method of claim 26, wherein determining indirect sales costs associated with the customer comprises determining indirect sales costs related to costs incurred before and after a particular sale to the customer.

14. A computer-readable medium having embodied thereon a non-transitory computer program configured to manage customer loss, the medium comprising one or more non-transitory code segments, that when executed, causes one or more processors to:

access customer information having multiple customer records, each customer record including multiple attribute values;

determine for each customer for which a record was accessed a churn likelihood representing the probability that the customer will be lost within a predetermined period of time;

compute, for each customer for which a record was accessed, a composite-customer-value index for the customer as a sum of (1) a product-cost profitability measure for the customer determined by subtracting product costs from the customer'"'"'s net sales, (2) a sales-cost profitability measure for the customer determined by subtracting, from the product-cost profitability measure, additional direct and indirect sales costs associated with the customer, and (3) a result of dividing the sales-cost profitability measure for the customer by the product-cost profitability measure for the customer, the composite-customer-value index representing the value of the customer to a business enterprise;

for each customer for which a record was accessed, normalize the composite-customer-value index for the customer and apply a first statistical coefficient to the composite-customer-value index for the customer to calculate a processed composite-customer-value index for the customer;

for each customer for which a record was accessed, apply a second statistical coefficient to the churn likelihood for the customer to calculate a processed churn likelihood index for the customer, the second statistical coefficient being different than the first statistical coefficient;

for each customer for which a record was accessed, compute a value-churn measure for the customer using the processed churn likelihood index for the customer and the processed composite-customer-value index for the customer;

access, for each of at least three likelihood-to-churn categories, a predetermined threshold that identifies a range of value-churn measures to be used for determining customers to be associated with a particular likelihood-to-churn category from among the at least three likelihood-to-churn categories;

analyze the determined value-churn measure for each customer with respect to the accessed predetermined thresholds;

based on the analysis, associate each customer with one of the at least three likelihood-to-churn categories, the association being based on the determined value-churn measure of a customer falling within the range of value-churn measures corresponding to the associated likelihood-to-churn category;

receive user input selecting, from among the at least three likelihood-to-churn categories, a likelihood-to-churn category to be used to cluster customers that are associated with the selected likelihood-to-churn category;

based on receiving user input selecting a likelihood-to-churn category to be used to cluster customers, cluster customers associated with the selected likelihood-to-churn category into multiple groups of customers having a shared characteristic that is different than likelihood-to-churn and value to the business enterprise; and

define, for each of the multiple groups of customers, a promotional campaign that is targeted to customers having a value-churn measure falling within the range of value-churn measures corresponding to the selected likelihood-to-churn category and that is targeted to the shared characteristic of the customers clustered into the corresponding group.

15. The medium of claim 14 wherein the one or more code segments are further configured to:

identify, for each customer represented by a customer record, an importance value that represents the value of the customer to a business enterprise; and

identify customer records that have both a high churn likelihood and a high importance value, wherein associating each customer in one of the likelihood-to-churn categories with one of several customer groups comprises associating each customer that has both a high churn likelihood and a high importance value into one of several customer groups.

16. The medium of claim 15 wherein the importance value comprises an importance value having at least two importance indicators.

17. The medium of claim 15 wherein the importance value comprises a profitability value that represents the contribution of the customer to the business enterprise.

18. The medium of claim 17 wherein the profitability value comprises a profitability value having 1) a product-cost value that represents a net sales-cost value arrived at by subtracting a sales deductions value from a gross sales value and 2) a sales-cost value arrived at by subtracting an additional cost value associated with selling to the customer from the product-cost value.

19. The medium of claim 15 wherein the one or more code segments are further configured to:

apply a first statistical weight to the product-cost value; and

apply a second statistical weight to the sales-cost value, wherein the profitability value comprises a profitability value based on the application of a first statistical weight to the product-cost value and the application of a second statistical weight to sales-cost value.

20. The medium of claim 19 wherein the first statistical weight and the second statistical weight are user-configurable.

21. A system for managing customer loss, the system comprising:

a processor connected to a storage device and one or more input/output devices, wherein the processor is configured to;

access customer information having multiple customer records, each customer record including multiple attribute values;

determine for each customer for which a record was accessed a churn likelihood representing the probability that the customer will be lost within a predetermined period of time;

compute, for each customer for which a record was accessed, a composite-customer-value index for the customer as a sum of (1) a product-cost profitability measure for the customer determined by subtracting product costs from the customer'"'"'s net sales, (2) a sales-cost profitability measure for the customer determined by subtracting, from the product-cost profitability measure, additional direct and indirect sales costs associated with the customer, and (3) a result of dividing the sales-cost profitability measure for the customer by the product-cost profitability measure for the customer, the composite-customer-value index representing the value of the customer to a business enterprise;

for each customer for which a record was accessed, normalize the composite-customer-value index for the customer and apply a first statistical coefficient to the composite-customer-value index for the customer to calculate a processed composite-customer-value index for the customer;

for each customer for which a record was accessed, apply a second statistical coefficient to the churn likelihood for the customer to calculate a processed churn likelihood index for the customer, the second statistical coefficient being different than the first statistical coefficient;

for each customer for which a record was accessed, compute a value-churn measure for the customer using the processed churn likelihood index for the customer and the processed composite-customer-value index for the customer;

access, for each of at least three likelihood-to-churn categories, a predetermined threshold that identifies a range of value-churn measures to be used for determining customers to be associated with a particular likelihood-to-churn category from among the at least three likelihood-to-churn categories;

analyze the determined value-churn measure for each customer with respect to the accessed predetermined thresholds;

based on the analysis, associate each customer with one of the at least three likelihood-to-churn categories, the association being based on the determined value-churn measure of a customer falling within the range of value-churn measures corresponding to the associated likelihood-to-churn category;

receive user input selecting, from among the at least three likelihood-to-churn categories, a likelihood-to-churn category to be used to cluster customers that are associated with the selected likelihood-to-churn category;

based on receiving user input selecting a likelihood-to-churn category to be used to cluster customers, cluster customers associated with the selected likelihood-to-churn category into multiple groups of customers having a shared characteristic that is different than likelihood-to-churn and value to the business enterprise; and

define, for each of the multiple groups of customers, a promotional campaign that is targeted to customers having a value-churn measure falling within the range of value-churn measures corresponding to the selected likelihood-to-churn category and that is targeted to the shared characteristic of the customers clustered into the corresponding group.

22. The system of claim 21 wherein the processor is further configured to:

identify, for each customer represented by a customer record, an importance value that represents the value of the customer to a business enterprise; and

identify customer records that have both a high churn likelihood and a high importance value, wherein associating each customer in one of the likelihood-to-churn categories with one of several customer groups comprises associating each customer that has both a high churn likelihood and a high importance value into one of several customer groups.