In the rapidly evolving world of telecommunications, delivering excellent customer support is essential for leading companies. A major Caribbean telecommunications conglomerate, with a strong presence in multiple countries and a commitment to innovation, aimed to improve its call center operations by adopting advanced data analytics. This article discusses how the integration of Generative AI (GenAI) transformed the customer support experience for this prominent telecom organization.
The Challenge: Turning Unstructured Data into Valuable Insights
The telecom giant faced a common challenge in their customer support centers: the inability to perform comprehensive statistical analysis on agent-customer interactions beyond basic metadata, such as call duration or response delay. Agents were initially asked to complete forms during live conversations, but this approach proved inconsistent and inefficient, as multitasking during calls affected the accuracy and completeness of the collected data.
To address this challenge, a cutting-edge solution was created using Generative AI to extract valuable insights from unstructured conversation data. This innovative tool processes phone conversation transcripts, identifies relevant information, and generates structured data for analysis. By categorizing each interaction and extracting features like intents, flags, assets, and resolutions, the tool greatly improved the consistency and quality of the data available for analytics.
A Complete Solution: AI-Powered Analytics, Orchestration, and Forensics
The AI-powered analytics solution includes three main parts: the analytics component, the orchestration process, and the Forensics tool.
- Analytics Component: This data extraction process manages client-agent interactions in text format, extracts target features from calls, and stores the results in a structured table. The tool segments interactions into the company’s main markets (mobile plans, home connection, TV) to enable more targeted extraction for each process.
- Orchestration Process: This component manages and troubleshoots the analytics system, scheduling it to run daily on the latest data. It guarantees smooth execution and easy monitoring of the entire process. Additionally, in case of an error or system downtime, it provides easy reprocessing of failed results and backfilling of skipped data.
- Forensics Tool (RAG): This user-friendly tool enables the client to ask natural-language questions about interactions. An LLM responds by retrieving summaries of relevant interaction data, offering a quick and simple way to assess the overall status or situation of interactions in a specific area.
The solution is built using various AWS services, including Step Functions and Glue Jobs for orchestration and data preparation, Lambda Functions for processing tasks, RDS and Glue Data Catalog for storing intermediate results, Athena for the final Analytics Table, and DynamoDB for process tracking.
Empowering Customer Support Using AI-Driven Insights
At Strata, we believe that even when we develop and provide the best tools to solve a problem or issue, it is when the client is able to utilize and manage these tools themselves, since they interact with their data and systems on a daily basis.
The implementation of this AI-powered analytics solution offered the telecommunications conglomerate several benefits:
- Structured Data for Deep Analysis: The analytics tool allowed the client’s Data Analytics team to perform statistical analysis of interactions between clients and agents in call centers, using structured data that was more insightful than previous sources.
- Automated Processing Pipeline: The orchestration component automated the processing pipeline, running daily with the latest input data to ensure efficient, current analytics.
- Initial Exploration and Overview: The Forensics tool provided a quick, simple way to explore interaction statuses, giving an initial overview of data for a specific subject.
Additionally, Knowledge Sharing sessions and detailed Runbooks were supplied to help the client use, run, and manage all parts of the tool independently.
In conclusion, the integrating Generative AI into this leading Caribbean telecommunications company’s customer support operations transformed how they analyze and understand agent-customer interactions. By turning unstructured conversation data into valuable insights, the AI-powered analytics solution empowers the telecom giant to keep improving customer support and stay competitive in the industry..
