Behavior Score Waveform Analysis for Call Center Coaching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current call center systems can determine agent behaviors but fail to identify the underlying events causing these behaviors, limiting the effectiveness of agent coaching by not providing actionable insights for improvement.
Innovation Solution
A method to analyze interactions between customers and agents, generating behavior score waveforms and identifying specific events associated with these behaviors, allowing for the calculation of behavior-event scores that indicate the relevance of events to agent behaviors, enabling customized and targeted coaching.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If behavior models are used to determine agent behaviors from interaction data, then behavior analysis capability is improved, but the ability to understand the cause of behaviors deteriorates because events surrounding the behavior are not identified
Solution Approach 1:
The system segments the interaction analysis into two distinct components: behavior detection (using behavior models to identify agent behaviors) and event detection (identifying specific events that occurred during interactions). This segmentation allows each component to specialize - behavior models focus on behavioral patterns while event detection focuses on identifying actionable events, thereby preserving cause-information without compromising behavior analysis precision
Solution Approach 2:
The system introduces an intermediary component that links behavior detection results with event detection results. This intermediary correlates behaviors with their surrounding events, enabling supervisors to understand not just what behaviors occurred but what events caused them. The intermediary preserves the causal relationship information that would otherwise be lost in pure behavior analysis
2Ease of operation
If coaching is provided based on behavior scores alone, then coaching can be delivered, but coaching effectiveness deteriorates due to lack of actionable insights from event identification
Solution Approach 1:
The system performs preliminary event identification and correlation with behaviors before coaching delivery. By pre-identifying which events caused which behaviors, the system prepares actionable insights in advance, making coaching more effective without increasing the complexity of the coaching delivery process itself
3Productivity
If random sampling of interactions is used for coaching analysis, then processing load is reduced, but coaching coverage and effectiveness deteriorate
Solution Approach 1:
The system extracts only the most relevant events from full interaction data using event detection algorithms. Instead of analyzing all interaction data in full detail, the system extracts key events that are most likely to be coachable, thereby maintaining high processing efficiency while ensuring comprehensive coaching coverage through targeted event selection
Data Source
AI summary
A method for providing coachable events for agents within a call center is provided. Behavior score waveforms for interactions and behaviors can be determined. Events can be identified in the behavior score waveforms within identified durations, and a relevancy of one or more events to one or more behaviors can be determined.