System, method, and computer program for deploying a prepackaged analytic intelligence module for a quote-to-cash application while protecting the privacy of customer data
The present disclosure describes a system, method, and computer program for creating and deploying prepackaged, predictive and prescriptive analytics modules for use with a quote-to-cash application. When deployed, the prepackaged analytics modules execute seamlessly, from the user's perspective, with the quote-to-cash application to provide predictions, recommendations, or other data-driven insights at one or more places in the quote-to-cash process. The modules are created by leveraging a pre-defined transactional data model that is native to the quote-to-cash application. A separate instance of the module is created for each customer that deploys a module. A customer's instance of the module is automatically retrained with the customer's own quote-to-cash data during the deployment phase by retrieving customer data matching metadata definitions predefined by the quote-to-cash application. Each of the prepackaged analytics modules is configured to work with at least one custom input attribute of a select data type and data distribution.
1. A method performed by a computer system comprising one or more computing devices for deploying a prepackaged analytics module for use with a quote-to-cash application executing on a remote system, the following steps performed by the one or more computing devices of the computer system:
maintaining on a server a library of prepackaged predictive and prescriptive analytics modules for a plurality of use cases in a quote-to-cash process, wherein:
the inputs to each of the prepackaged analytics modules are attributes of data objects that are native to a data model used by the quote-to-cash application,
each of the input attributes, as well as outcome data used to train the prepackaged analytics modules, are associated with a metadata definition that corresponds to metadata definitions in the data model used by the quote-to-cash application, and
the prepackaged analytics modules in the library are available for deployment by a plurality of different customers of the quote-to-cash application and were created without human access to said customers' data;
receiving a request from a select customer to deploy a prepackaged analytics module for a use case to a remote server;
identifying a prepackaged analytics module in the library for the use case;
creating a customer-specific instance of the identified prepackaged analytics module on the remote server;
identifying a first subset of the select customer's quote-to-cash data based on the metadata definitions for input attributes and outcome data of the identified prepackaged analytics module;
retraining only the customer-specific instance of the identified prepackaged analytics module using the first subset of the select customer's quote-to-cash data;
identifying a second subset of the select customer's quote-to-cash data that matches the metadata definitions for the input attributes and outcome data of the identified prepackaged analytics module;
validating the retrained, customer-specific instance of the identified prepackaged analytics module using the second subset of quote-to-cash data, wherein the validating step comprises applying the inputs of the second subset of quote-to-cash data to the retrained, customer-specific instance of the identified prepackaged analytics module to determine how reliably it predicts the corresponding outcome data in the second subset of quote-to-cash data;
determining whether the retrained, customer-specific instance of the identified prepackaged analytics module satisfies an accuracy threshold; and
in response to the retrained, customer-specific instance of the identified prepackaged analytics module satisfying the accuracy threshold, deploying the retrained, customer-specific instance of the identified prepackaged analytics module for use by only the select customer with the quote-to-cash application.
2. The method of claim 1 , wherein a plurality of prepackaged analytics modules are identified for the use case, wherein a customer-specific instance of each of the identified prepackaged analytics modules is created, retrained, and validated, and wherein the most accurate of the instances is identified and deployed for use by the select customer with the quote-to-cash application.
3. The method of claim 1 , wherein each of the prepackaged analytics modules is configured to work with at least one custom input attribute of a select data type and data distribution, and wherein the method further comprises:
in response to the retrained, customer-specific instance of the identified prepackaged analytics module not satisfying the accuracy threshold, determining whether the customer quote-to-cash data includes an attribute matching the data type and data distribution associated with the custom input attribute;
in response to determining that the customer quote-to-cash data includes an attribute matching the data type and data distribution of the custom input attribute, retraining and revalidating the customer-specific instance of the identified prepackaged analytics module;
determining whether the customer-specific instance of the identified prepackaged analytics module with the custom attribute satisfies the accuracy threshold; and
in response to the customer-specific instance of the identified prepackaged analytics module with the custom attribute satisfying the accuracy threshold, deploying said customer-specific instance of the identified prepackaged analytics module for use by only the select customer with the quote-to-cash application.
4. The method of claim 1 , wherein the library includes sets of prepackaged analytics modules for each of a plurality of industries and each of a plurality of use cases within each industry.
5. A non-transitory computer-readable medium comprising a computer program, that, when executed by a computer system, enables the computer system to perform the following method for deploying a prepackaged, predictive and prescriptive analytics module for use with a quote-to-cash application executing on a remote system, the steps comprising:
maintaining a library of prepackaged analytics modules for a plurality of use cases in a quote-to-cash process, wherein:
the inputs to each of the prepackaged analytics modules are attributes of data objects that are native to a data model used by the quote-to-cash application,
each of the input attributes, as well as outcome data used to train the prepackaged analytics modules, are associated with a metadata definition that corresponds to metadata definitions in the data model used by the quote-to-cash application, and
the prepackaged analytics modules in the library are available for deployment by a plurality of different customers of the quote-to-cash application and were created without access to said customers' data;
receiving a request from a select customer to deploy a prepackaged analytics module for a use case;
identifying a prepackaged analytics module in the library for the use case;
creating a customer-specific instance of the identified prepackaged analytics module;
identifying a first subset of the select customer's quote-to-cash data based on the metadata definitions for input attributes and outcome data of the identified prepackaged analytics module;
retraining only the customer-specific instance of the identified prepackaged analytics module using the first subset of the select customer's quote-to-cash data;
identifying a second subset of the select customer's quote-to-cash data that matches the metadata definitions for the input attributes and outcome data of the identified prepackaged analytics module;
validating the retrained, customer-specific instance of the identified prepackaged analytics module using the second subset of quote-to-cash data, wherein the validating step comprises applying the inputs of the second subset of quote-to-cash data to the retrained, customer-specific instance of the identified prepackaged analytics module to determine how reliably it predicts the corresponding outcome data in the second subset of quote-to-cash data;
determining whether the retrained, customer-specific instance of the identified prepackaged analytics module satisfies an accuracy threshold; and
in response to the retrained, customer-specific instance of the identified prepackaged analytics module satisfying the accuracy threshold, deploying the retrained, customer-specific instance of the identified prepackaged analytics module for use by only the select customer with the quote-to-cash application.
6. The non-transitory computer-readable medium of claim 5 , wherein a plurality of prepackaged analytics modules are identified for the use case, wherein a customer-specific instance of each of the identified prepackaged analytics modules is created, retrained, and validated, and wherein the most accurate of the instances is identified and deployed for use by the select customer in the quote-to-cash application.
7. The non-transitory computer-readable medium of claim 5 , wherein each of the prepackaged analytics modules is configured to work with at least one custom input attribute of a select data type and data distribution, and wherein the method further comprises:
in response to the retrained, customer-specific instance of the identified prepackaged analytics module not satisfying the accuracy threshold, determining whether the customer quote-to-cash data includes an attribute matching the data type and data distribution associated with the custom input attribute;
in response to determining that the customer quote-to-cash data includes an attribute matching the data type and data distribution of the custom input attribute, retraining and re-validating the customer-specific instance of the identified prepackaged analytics module;
determining whether the customer-specific instance of the identified prepackaged analytics module with the custom attribute satisfies the accuracy threshold; and
in response to the customer-specific instance of the identified prepackaged analytics module with the custom attribute satisfying the accuracy threshold, deploying said customer-specific instance of the identified prepackaged analytics module for use by only the select customer with the quote-to-cash application.
8. The non-transitory computer-readable medium of claim 5 , wherein the library includes sets of prepackaged analytics modules for each of a plurality of industries and each of a plurality of use cases within each industry.
9. The non-transitory computer-readable medium of claim 5 , wherein the library is maintained on a first system with no access to the select customer's quote-to-cash data, and wherein the customer-specific instance of the identified prepackaged analytics module is retrained and deployed on a second system with access to the select customer's quote-to-cash data.
10. A computer system for deploying a prepackaged,
predictive and prescriptive analytics module for use with a quote-to-cash application executing, the system comprising:
one or more processors;
one or more memory units coupled to the one or more processors, wherein the one or more memory units store instructions that, when executed by the one or more processors, cause the system to perform the operations of:
maintaining a library of prepackaged analytics modules for a plurality of use cases in a quote-to-cash process, wherein:
the inputs to each of the prepackaged analytics modules are attributes of data objects that are native to a data model used by the quote-to-cash application,
each of the input attributes, as well as outcome data used to train the prepackaged analytics modules, are associated with a metadata definition that corresponds to metadata definitions in the data model used by the quote-to-cash application, and
the prepackaged analytics modules in the library are available for deployment by a plurality of different customers of the quote-to-cash application and were created without direct human access to said customers' data;
receiving a request from a select customer to deploy a prepackaged analytics module for a use case;
identifying a prepackaged analytics module in the library for the use case;
creating a customer-specific instance of the identified prepackaged analytics module;
identifying a first subset of the select customer's quote-to-cash data based on the metadata definitions for input attributes and outcome data of the identified prepackaged analytics module;
retraining only the customer-specific instance of the identified prepackaged analytics module using the first subset of the select customer's quote-to-cash data;
identifying a second subset of the select customer's quote-to-cash data that matches the metadata definitions for the input attributes and outcome data of the identified prepackaged analytics module;
validating the retrained, customer-specific instance of the identified prepackaged analytics module using the second subset of quote-to-cash data, wherein the validating step comprises applying the inputs of the second subset of quote-to-cash data to the retrained, customer-specific instance of the identified prepackaged analytics module to determine how reliably it predicts the corresponding outcome data in the second subset of quote-to-cash data;
determining whether the retrained, customer-specific instance of the identified prepackaged analytics module satisfies an accuracy threshold; and
in response to the retrained, customer-specific instance of the identified prepackaged analytics module satisfying the accuracy threshold, deploying the retrained, customer-specific instance of the identified prepackaged analytics module for use by only the select customer with the quote-to-cash application.
11. The system of claim 10 , wherein a plurality of prepackaged analytics modules are identified for the use case, wherein a customer-specific instance of each of the identified prepackaged analytics modules is created, retrained, and validated, and wherein the most accurate of the instances is identified and deployed for use by the select customer with the quote-to-cash application.
12. The system of claim 10 , wherein each of the prepackaged analytics modules is configured to work with at least one custom input attribute of a select data type and data distribution, and wherein the method further comprises:
in response to the retrained, customer-specific instance of the identified prepackaged analytics module not satisfying the accuracy threshold, determining whether the customer quote-to-cash data includes an attribute matching the data type and data distribution associated with the custom input attribute;
in response to determining that the customer quote-to-cash data includes an attribute matching the data type and data distribution of the custom input attribute, retraining and re-validating the customer-specific instance of the identified prepackaged analytics module;
determining whether the customer-specific instance of the identified prepackaged analytics module with the custom attribute satisfies the accuracy threshold; and
in response to the customer-specific instance of the identified prepackaged analytics module with the custom attribute satisfying the accuracy threshold, deploying said customer-specific instance of the identified prepackaged analytics module for use by only the select customer with the quote-to-cash application.
13. The system of claim 10 , wherein the library includes sets of prepackaged analytics modules for each of a plurality of industries and each of a plurality of use cases within each industry.
14. The system of claim 10 , wherein the library is maintained on a first system with no direct human access to the select customer's quote-to-cash process data, and wherein the customer-specific instance of the identified prepackaged analytics module is retrained and deployed on a second system with access to the select customer's quote-to-cash process data.
15. The method of claim 1 , wherein recommendations or other data-driven insights generated by a deployed customer-specific instance of an identified prepackaged analytics module are integrated into a customer-specific instance of the quote-to-cash application and displayed in a user interface generated by the customer-specific instance of the quote-to-cash application.
16. The non-transitory computer-readable medium of claim 5 , wherein recommendations or other data-driven insights generated by a deployed customer-specific instance of an identified prepackaged analytics module are integrated into a customer-specific instance of the quote-to-cash application and displayed in a user interface generated by the customer-specific instance of the quote-to-cash application.
17. The system of claim 10 , wherein recommendations or other data-driven insights generated by a deployed customer-specific instance of an identified prepackaged analytics module are integrated into a customer-specific instance of the quote-to-cash application and displayed in a user interface generated by the customer-specific instance of the quote-to-cash application.