Enbridge’s customer service team fields thousands of support calls every month. These interactions are rich with insights — but until recently, most of that information was locked inside unstructured call transcripts. Challenges included:

• Limited visibility into common call drivers and customer pain points
• Difficulty identifying emerging issues in real time
• Inconsistent response quality across agents and regions
• A lack of data to guide improvements in call scripts or self-service options

Without systematic analysis, Enbridge was missing opportunities to improve customer experience and operational efficiency.

CVM partnered with Enbridge to analyze customer service call transcripts using natural language processing (NLP). Our goal: extract actionable insights from thousands of conversations — and help the client improve how they support customers.

Key elements of the solution included:

• Data Preparation: Processed and cleaned large volumes of call transcript data to enable structured analysis.
• Topic Modeling and Classification: Used NLP techniques to identify recurring themes, categorize call types, and surface emerging trends.
• Sentiment and Escalation Analysis: Flagged negative sentiment, complex issues, and escalation-prone topics to guide training and triage.
• Script Optimization: Identified opportunities to refine agent response scripts based on high-performing interactions and common questions.
• Insight Reporting: Delivered findings through dashboards and briefing materials tailored to frontline managers and CX leaders.

The system was built using Python and integrated into Enbridge’s analytics environment for ongoing use.

• Deeper Customer Insight: Enbridge gained a clearer understanding of why customers were calling — and how those reasons were evolving over time.
• Improved Agent Performance: Script enhancements led to more consistent, effective responses across the customer service team.
• Faster Issue Detection: Early warning signals for emerging call types allowed proactive adjustments to operations and messaging.
• Operational Efficiency: Reduced call handling time and repeat contact rates by better aligning support content with customer needs.

• NLP Makes Conversations Actionable: Analyzing unstructured call data helps organizations unlock new insights and improve service quality.
• Small Changes Have Big Impacts: Optimizing call scripts and workflows based on real-world data can drive meaningful performance gains.
• Insight Belongs on the Frontline: Delivering findings to the people closest to the work — not just senior leaders — maximized adoption and impact.