Using cognitive computing to improve relationship pricing
Techniques are disclosed utilizing cognitive computing to assess customer value and provide specific promotional campaigns based upon this assessed value. Users may opt in to a rewards program. With user permission or affirmative consent, user behavioral data may be monitored that may be relevant to the user's relationship as a customer with a particular business and may include various indications of the users' behaviors, actions, and/or preferences. This data may be stored as part of each user's behavioral profile, the contents of which may be analyzed to determine which customers are more profitable to the business than others. Each user may be assigned a customer value indicative of his or her individual profitability, which may be used to provide specific promotional campaigns in an attempt to maintain the more profitable customers and to improve the profitability of others.
1. A computer-implemented method, comprising:
training, by one or more processors, a cognitive computing machine learning model using historical data associated with:
a set of behavioral traits associated with a set of customers; and
a service provider value metric associated with a service provider,
wherein the training includes determining one or more correlations between one or more behavioral traits in the set of behavioral traits and the service provider value metric;
receiving, by the one or more processors, user behavioral data associated with a user;
generating, by the one or more processors, a user behavioral profile associated with the user and based upon the user behavioral data, the user behavioral profile indicating one or more user behavioral traits of the user;
determining, by the one or more processors, a customer value of the user to the service provider, based at least in part on at least one correlation, of the one or more correlations, between with the one or more user behavioral traits and the service provider value metric;
ranking, by the one or more processors, the customer value at a position within a scale of customer values ranging between a first level of value to the service provider and a second level of value to the service provider, the first level of value being less than the second level of value; and
associating, by the one or more processors, the user behavioral profile to a promotional campaign of the service provider, wherein the promotional campaign is associated with the position of the customer value within the scale of customer values.
2. The computer-implemented method of claim 1 , wherein determining the customer value comprises:
assigning, by the one or more processors, weights to individual user behavioral traits of the one or more user behavioral traits, based on the one or more correlations determined by training the cognitive computing machine learning model;
assigning, by the one or more processors, numeric values to the individual user behavioral traits of the one or more user behavioral traits, based at least in part on indications in the user behavioral profile of how often the user exhibits the individual user behavioral traits; and
determining, by the one or more processors, the customer value as a weighted average based on the weights and the numeric values assigned to the one or more user behavioral traits.
3. The computer-implemented method of claim 1 , wherein determining the customer value comprises:
accessing, by the one or more processors, a database of tiered reference behavioral profiles indicating sets of user behavioral traits associated with customers of different tiers of known customer values;
identifying, by the one or more processors, a matching reference behavioral profile from among the tiered reference behavioral profiles that has a highest number of user behavioral traits in common with the user behavioral profile; and
determining, by the one or more processors, the customer value based at least in part on a known customer value associated with the matching reference behavioral profile.
4. The computer-implemented method of claim 1 , wherein the scale of customer values is subdivided into a plurality of customer value ranges, and wherein determining the customer value comprises:
determining, by the one or more processors, a plurality of outputs from a predictive modeling function, each of the plurality of outputs indicating a different statistical likelihood of a future customer value associated with the user being within a corresponding customer value range of the plurality of customer value ranges; and
determining, by the one or more processors, the customer value based at least in part on an average of a set of customer values within a particular one of the plurality of customer value ranges that has a highest statistical likelihood associated with the future customer value from among the different statistical likelihoods.
5. The computer-implemented method of claim 1 , wherein the one or more user behavioral traits indicated in the user behavioral profile includes one or more of (i) spending data associated with the user; (ii) a preference of the user to receive information in a paperless manner; (iii) social networking data associated with the user; (iv) tracked location data associated with the user; (v) a likelihood that the user will terminate an ongoing relationship with the service provider; and (vi) a length of the ongoing relationship between the user and the service provider.
6. The computer-implemented method of claim 1 , wherein linking the user to the promotional campaign comprises one or more of:
adjusting, by the one or more processors, rewards program incentives for the user to provide user-specific reward points, the rewards program incentives being associated with a rewards program supported by the service provider; and
causing, by the one or more processors, the user to be provided with user-specific discounted pricing for one or more services provided by the service provider.
7. A system, comprising:
one or more processors;
memory storing computer-executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
training a cognitive computing machine learning model using historical behavior data associated with:
a set of behavioral traits associated with a set of customers; and
a service provider value metric associated with a service provider,
wherein the training includes determining one or more correlations between one or more behavioral traits in the set of behavioral traits and the service provider value metric;
receiving user behavioral data associated with a user;
generating a user behavioral profile associated with the user and based upon the user behavioral data, the user behavioral profile indicating one or more user behavioral traits of the user;
determining a customer value of the user to the service provider, based at least in part on at least one correlation, of the one or more correlations, between the one or more user behavioral traits and the service provider value metric;
ranking the customer value at a position within a scale of customer values ranging between a first level of value to the service provider and a second level of value to the service provider, the first level of value being less than the second level of value; and
associating the user behavioral profile to a promotional campaign of the service provider, wherein the promotional campaign is associated with the position of the customer value within the scale of customer values.
8. The system of claim 7 , wherein determining the customer value comprises:
assigning weights to individual user behavioral traits of the one or more user behavioral traits, based on the one or more correlations determined by training the cognitive computing machine learning model;
assigning numeric values to the individual user behavioral traits of the one or more user behavioral traits, based at least in part on indications in the user behavioral profile of how often the user exhibits the individual user behavioral traits; and
determining the customer value as a weighted average based on the weights and the numeric values assigned to the one or more user behavioral traits.
9. The system of claim 7 , wherein determining the customer value comprises:
accessing a database of tiered reference behavioral profiles indicating sets of user behavioral traits associated with customers of different tiers of known customer values;
identifying a matching reference behavioral profile from among the tiered reference behavioral profiles that has a highest number of user behavioral traits in common with the user behavioral profile; and
determining the customer value based at least in part on a known customer value associated with the matching reference behavioral profile.
10. The system of claim 7 , wherein the scale of customer values is subdivided into a plurality of customer value ranges, and wherein determining the customer value comprises:
determining a plurality of outputs from a predictive modeling function, each of the plurality of outputs indicating a different statistical likelihood of a future customer value associated with the user being within a corresponding customer value range of the plurality of customer value ranges; and
determining the customer value based at least in part on an average of a set of customer values within a particular one of the plurality of customer value ranges that has a highest statistical likelihood associated with the future customer value from among the different statistical likelihoods.
11. The system of claim 7 , wherein the one or more user behavioral traits indicated in the user behavioral profile includes one or more of (i) spending data associated with the user; (ii) a preference of the user to receive information in a paperless manner; (iii) social networking data associated with the user; (iv) tracked location data associated with the user; (v) a likelihood that the user will terminate an ongoing relationship with the service provider; and (vi) a length of the ongoing relationship between the user and the service provider.
12. The system of claim 7 , wherein linking the user to the promotional campaign comprises one or more of:
adjusting rewards program incentives for the user to provide user-specific reward points, the rewards program incentives being associated with a rewards program supported by the service provider; and
causing the user to be provided with user-specific discounted pricing for one or more services provided by the service provider.
13. A non-transitory, tangible computer-readable medium storing instructions thereon that, when executed by one or more processors, cause the one or more processors to:
train a cognitive computing machine learning model using historical behavior data associated with:
a set of behavioral traits associated with a set of customers; and
a service provider value metric associated with a service provider,
wherein the training includes determining one or more correlations between one or more behavioral traits in the set of behavioral traits and service provider value metric;
receive user behavioral data associated with a user;
generate a user behavioral profile associated with the user and based upon the user behavioral data, the user behavioral profile indicating one or more user behavioral traits of to the user;
determine a customer value of the user to the service provider, based at least in part on at least one correlation, of the one or more correlations, between the one or more user behavioral traits and the service provider value metric;
rank the customer value at a position within a scale of customer values ranging between a first level of value to the service provider and a second level of value to the service provider, the first level of value being less than the second level of value; and
associate the user behavioral profile to a promotional campaign of the service provider, wherein the promotional campaign is associated with the position of the customer value within the scale of customer values.
14. The non-transitory, tangible computer-readable medium of claim 13 , wherein the instructions cause the one or more processors to determine the customer value by:
assigning weights to individual user behavioral traits of the one or more user behavioral traits, based on the one or more correlations determined by training the cognitive computing machine learning model;
assigning numeric values to the individual user behavioral traits of the one or more user behavioral traits, based at least in part on indications in the user behavioral profile of how often the user exhibits the individual user behavioral traits; and
determining the customer value as a weighted average based on the weights and the numeric values assigned to the one or more user behavioral traits.
15. The non-transitory, tangible computer-readable medium of claim 13 , wherein the instructions cause the one or more processors to determine the customer value by:
accessing a database of tiered reference behavioral profiles indicating sets of user behavioral traits associated with customers of different tiers of known customer values;
identifying a matching reference behavioral profile from among the tiered reference behavioral profiles that has a highest number of user behavioral traits in common with the user behavioral profile; and
determining the customer value based at least in part on known customer value associated with the matching reference behavioral profile.
16. The non-transitory, tangible computer-readable medium of claim 13 , wherein the scale of customer values is subdivided into a plurality of customer value ranges, and wherein the instructions cause the one or more processors to determine the customer value by:
determining a plurality of outputs from a predictive modeling function, each of the plurality of outputs indicating a different statistical likelihood of a future customer value associated with the user being within a corresponding customer value range of the plurality of customer value ranges; and
determining the customer value based at least in part on an average of a set of customer values within a particular one of the plurality of customer value ranges that has a highest statistical likelihood associated with the future customer value from among the different statistical likelihoods.
17. The non-transitory, tangible computer-readable medium of claim 13 , wherein the one or more user behavioral traits indicated in the user behavioral profile includes one or more of (i) spending data associated with the user; (ii) a preference of the user to receive information in a paperless manner; (iii) social networking data associated with the user; (iv) tracked location data associated with the user; (v) a likelihood that the user will terminate an ongoing relationship with the service provider; and (vi) a length of the ongoing relationship between the user and the service provider.
18. The non-transitory, tangible computer-readable medium of claim 13 , wherein the instructions cause the one or more processors to link the user to the promotional campaign by:
adjusting rewards program incentives for the user to provide user-specific reward points, the rewards program incentives being associated with a rewards program supported by the service provider; and
causing the user to be provided with user-specific discounted pricing for one or more services provided by the service provider.
19. The computer-implemented method of claim 4 , wherein determining the customer value further comprises averaging, by the one or more processors, (i) the average of the set of customer values within the particular one of the plurality of customer value ranges with (ii) a current customer value associated with the user.
20. The system of claim 10 , wherein determining the customer value further comprises averaging (i) the average of the set of customer values within the particular one of the plurality of customer value ranges with (ii) a current customer value associated with the user.
21. The non-transitory, tangible computer-readable medium of claim 16 , wherein the instructions cause the one or more processors to further determine the customer value by averaging (i) the average of the set of customer values within the particular one of the plurality of customer value ranges with (ii) a current customer value associated with the user.