Real-Time Phrase Selection for Customer Support Agents
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Customer support agents face challenges in promptly identifying effective phrases to direct conversations due to high call traffic and unpredictable conversation contexts, making it difficult to select appropriate responses from voluminous scripted resources.
Innovation Solution
A real-time conversation monitoring system using machine learning models to detect conversation attributes and identify desired outcomes, presenting phrases with confidence scores based on past interactions to guide agents in achieving desired conversation outcomes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If agents use voluminous scripted resources to respond to customers, then they have a variety of responses available, but it becomes difficult to quickly identify appropriate and effective phrases
Solution Approach 1:
The system provides real-time feedback to agents by analyzing conversation attributes and recommending specific phrases from the script library. The machine learning model processes ongoing conversations and delivers actionable recommendations, enabling agents to quickly identify effective phrases without manually searching through voluminous scripted resources.
Solution Approach 2:
The patent introduces an intermediary system (machine learning model and conversation analysis system) that acts as a mediator between the agent and the scripted resources. This intermediary automatically analyzes conversation context and matches appropriate phrases, eliminating the need for agents to manually search through scripts while ensuring versatile and context-appropriate responses.
2Ease of operation
If agents rely on prescribed scripted responses, then they have guidance available, but the scripts may be inapt to the specific situation or external circumstances
Solution Approach 1:
The system dynamically adapts phrase recommendations based on real-time conversation attributes and context. Rather than providing static scripted responses, the machine learning model continuously analyzes conversation flow, detected attributes, and external circumstances to recommend phrases that are specifically apt for the current situation, making the guidance both easy to access and highly adaptable.
Solution Approach 2:
The patent applies local quality by tailoring phrase recommendations to specific conversation contexts and attributes. The system identifies relevant conversation characteristics (e.g., customer sentiment, issue type, conversation stage) and recommends phrases that are locally optimized for each specific situation rather than providing generic scripted responses.
3Measurement precision
If agents manually evaluate response effectiveness, then they can assess appropriateness, but they cannot evaluate quickly enough to match conversation pace
Solution Approach 1:
The patent replaces the mechanical system of manual human evaluation with an automated machine learning-based evaluation system. The system automatically analyzes conversation attributes, predicts outcome effectiveness, and provides real-time recommendations at speeds that match conversation pace, while maintaining or improving evaluation accuracy through data-driven insights.
Solution Approach 2:
The system enables self-service evaluation by automatically assessing phrase effectiveness and providing recommendations without requiring agent intervention. The machine learning model continuously monitors conversations, evaluates potential phrase effectiveness, and delivers recommendations autonomously, allowing agents to maintain evaluation accuracy while keeping hands free for customer interaction.
Data Source
AI summary
A conversation may be monitored in real time using a trained machine learning model to identify a desired outcome of a conversation and generate one or more phrases for accomplishing the desired outcome. A confidence score may also be determined for one or more phrases that indicates a likelihood that the one or more phrases may help accomplish the desired outcome of the conversation. In some examples, a confidence score may be based on whether an agent, a caller, or both responded unfavorably to a similar phrase used previously in another conversation. In other examples, a confidence score corresponding to one or more phrases may be based on whether a prior conversation in which one or more similar phrases was used resulted in the desired outcome being accomplished.