Boosting Customer Satisfaction via Practical Language Processing on Bank Call Centre Customer Dialog Data
Masters of Technology in Artificial Intelligence (National University of Singapore - ISS)
Practice Module 4
The 3rd party agencies claim to have highly effective proprietary methods which the bank does not have access to. Within the bank, the categorization and tabulation of customer satisfaction
levels is performed manually, with results documented on Excel spreadsheets. This laborious process is error-prone and hinders the generation of timely customer satisfaction updates,
typically limited to a quarterly frequency.
Hence, management would like to try using Practical Language Processing (PLP) techniques on customer dialog data to:
- Automate the manual process of monitoring customer satisfaction levels.
- Identify areas where the bank is doing well or not, with system recommended prioritisation.
- Validate the accuracy of the existing customer satisfaction monitoring services.
-
By implementing this PLP Analysis System, the bank will be able to take a data-driven approach to gain deep insights into their customer interactions, allowing them to proactively address emerging issues and adapt quickly to changing customer needs and expectations.
The system identifies topics from customer dialog, identifies the product and usage areas the bank is doing well (or not), and prioritises them based on a relevant importance metric so that the bank can make a data-driven decision on intervention initiatives. In addition to that, from analysis of customer dialog, the application also provides bank wide metrics like Net Promoter Score (NPS) and Customer Satisfaction Ratio (CSAT) for the bank to have an overall sense of their standing in their customers' eyes.
From Example Dialog like this ...
the Dialog Analysis Application uses 4 different techniques to
- Identify key customer topics
- Ranks each topic by customer sentiment and relevant importance metrics
- Calculates the bank's overall NPS and CSAT scores.
