System and method for managing routing of customer calls to agents
摘要
說明書
In the system and method of the present disclosure, for each identified customer predictive modeling module 410 selects among two or more predictive models by selecting a predictive model for which a set of enterprise customer data for the identified customer has the highest expected impact on likelihood of a respective business outcome associated with that predictive model. In an embodiment, each of the plurality of predictive models is configured to determine a business outcome signal representative of one of more of likelihood of accepting an offer to purchase a product, and the selected predictive model is the one of the plurality of predictive models for which the set of enterprise customer data has a highest importance in determining the respective business outcome signal. In an embodiment, analytical engine 404 analyzes the set of enterprise customer data retrieved and selects either a first predictive model that includes regression model #1 412 and tree-based model #1 414 (collectively called predictive model #1), or a second predictive model that includes regression model #2 416 and tree-based model #2 418 (collectively called predictive model #2). In other embodiments, predictive modeling module 410 can include more than two predictive models, and analytical engine 404 can select combinations of component models (e.g., regression model and tree-based model) other than predetermined pairings of these component models. The selection of predictive models based upon consistency of the selected predictive model as a modeling target with a set of enterprise customer data retrieved for an identified customer improves reliability of value-based classification of identified customer for call routing. In addition, predictive modeling module 410 may build stronger predictive models using enterprise customer data of the sponsoring organization.
