Predictive Routing for Service Sessions Using Activity Records
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional service environments, such as call centers, often rely on manual or automatic call routing systems that are prone to errors, leading to misrouting of service requests and subsequent rerouting operations, which consume processing power and resources.
Innovation Solution
Implementing a predictive routing system that detects user actions in applications to determine the category of service requests and automatically routes them to suitable service representatives, bypassing traditional IVR interfaces and reducing errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional manual or automatic call routing systems are used, then service requests can be routed to service representatives, but routing errors occur leading to misrouting and subsequent rerouting operations
Solution Approach 1:
The system performs preliminary detection of user actions in applications before routing occurs. By detecting actions (such as viewing account information, making transactions) and storing them in an activity data store, the system prepares category information in advance. When a service request is received, the system retrieves pre-stored activity records and determines the category before routing, eliminating the need for error recovery operations and improving both reliability and productivity.
2Reliability
If traditional routing systems with IVR interfaces are used, then service requests can be handled, but processing power and resources are consumed for error recovery and rerouting
Solution Approach 1:
The system performs preliminary detection of user actions and stores category information in an activity data store before routing occurs. By preparing the category determination in advance based on detected actions, the system eliminates the need for resource-intensive error recovery and rerouting operations, thereby reducing processing power consumption while maintaining high routing accuracy.
Solution Approach 2:
The system extracts and stores only the relevant category information from user actions in the activity data store, rather than processing all possible routing scenarios. This extraction of essential information enables accurate routing without requiring extensive processing power for error recovery, as the category is already determined from the extracted action data.
3Reliability
If predictive routing based on detected user actions is implemented, then routing accuracy is improved, but the system complexity increases due to action detection and activity record storage
Solution Approach 1:
The system segments the routing process into distinct functional components: action detection module, activity data store, category determination module, and routing module. By segmenting these functions, the system manages complexity through modular architecture while achieving improved routing accuracy. Each component handles a specific aspect of the predictive routing process independently.
Solution Approach 2:
The activity data store serves as an intermediary between action detection and routing decisions. It stores category information derived from user actions, acting as a mediator that simplifies the relationship between detection and routing functions. This intermediary structure reduces direct system complexity by providing a clear data layer that connects detection results to routing actions.
Data Source
AI summary
Techniques are described for routing service requests. Actions performed by an individual in an application are detected and used to determine a category of product or service that the individual is interested in. An activity record can be stored in an activity data store, the activity record identifying the individual and the category of interest determined for the individual. On subsequently receiving a service request from the individual, to an environment such as a call center, the activity record can be accessed and used to predict the individual's service need as corresponding to the category determined based on their previous actions. The service request is then automatically routed to a service representative who is suited to handle requests in that particular category. In some implementations, the category can be predicted and/or inferred based on the detected user activities in particular portions of the application.