Customer Representative Rating via Chat Transcript Success Vectors

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Customer representatives in live chat interfaces often lack the necessary skills to address diverse customer questions efficiently, leading to time-consuming searches for answers and potential inability to resolve issues effectively, with skill matching depending on business priorities.

Innovation Solution

A method that includes receiving and analyzing chat transcripts to generate multi-dimensional success vectors, rating customer representatives based on their responses, and matching customers with the most qualified representatives using machine learning models trained on historical data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If customer representatives are educated on only a portion of questions, then training time and resources are reduced, but the ability to effectively assist customers across diverse topics deteriorates

Engineering Contradiction:

Improvetraining efficiencyVSAvoidskill coverage

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system pre-processes chat transcripts to extract triplets and generate success vectors in advance, creating a ready-to-use knowledge base that enables rapid skill assessment and matching without requiring extensive real-time training of representatives

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

A machine learning model acts as an intermediary between customer representatives and customer questions, automatically matching representatives to customers based on skill vectors derived from historical chat data, eliminating the need for representatives to have exhaustive knowledge of all topics

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If customer representatives search for appropriate answers, then comprehensive coverage of topics is improved, but response time increases

Engineering Contradiction:

Improveanswer accuracyVSAvoidresponse time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system pre-generates success vectors from historical chat transcripts and stores them in a database, so that during live chats, the machine learning model can quickly retrieve and match appropriate representatives without time-consuming searches

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The manual search process is replaced with an automated machine learning-based matching system that automatically retrieves relevant success vectors and identifies suitable representatives, eliminating the need for representatives to manually search for answers

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If customers are transferred to different representatives, then skill matching is improved, but customer wait time and resolution time increase

Engineering Contradiction:

Improveskill matching accuracyVSAvoidresolution time

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system pre-calculates success vectors from historical data and maintains them in a database, enabling the machine learning model to perform rapid skill-based matching during live chats without requiring time-consuming transfers or reassignments

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system uses historical chat transcript outcomes to generate success vectors that provide feedback on representative performance, continuously improving matching accuracy by learning from past successful interactions

Inventive Principle:

Principle #23Feedback

4Measurement precision

If multi-dimensional success vectors are generated and normalized, then rating precision is improved, but computational complexity increases

Engineering Contradiction:

Improverating accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments chat transcripts into discrete triplets (customer question, representative response, outcome) and generates success vectors for each triplet independently, then aggregates them to create comprehensive ratings, breaking down the complex task into manageable parts

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS11227250B2Rating customer representatives based on past chat transcripts

Publication Date: 2022.01.18 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A method, computer system, and a computer program product for customer representative ratings is provided. The present invention may include receiving a chat transcript with one or more tagged triplets and one or more multi-dimensional success vectors. The present invention may include aggregating the one or more multi-dimensional success vectors. The present invention may include receiving at least one business priority. The present invention may include applying at least one filter to the one or more multi-dimensional success vectors. The present invention may include normalizing the one or more multi-dimensional success vectors based on the at least one applied filter. The present invention may include obtaining a rating.