Frontier Communications Leverages Amazon Bedrock for Enhanced Customer Insights Through Call Transcript Analysis

Frontier Communications uses Amazon Bedrock to analyze call transcripts and improve customer insights. The company handles 125,000 calls per month, with 250 million words spoken by agents. Natural Language Processing (NLP) was initially used to analyze the transcripts, but was laborious and inflexible. Frontier turned to large language models (LLMs) to classify call intent into 50 mutually exclusive categories, allowing for the analysis of derivative questions such as customer cancellation reasons and agent performance. The use of LLMs enables Frontier to provide actionable insights and improve customer retention and loyalty.