success story

Over 30% improvement for chatbot containment rates at a leading American bank

Using Virtusa’s Effective Virtual Agents Solution

The Challenge

One of the world’s largest global banks implemented a conversational AI solution to enhance customer service. To help further automate the consumer banking business functions, they developed over 200 actions for the chatbot feature. Upon analyzing the solution, they discovered the chatbot had less than a 15% containment rate. Customers were abandoning the chats and eventually stopped using the LiveChat portal.

The Solution

Effective Virtual Agents Solution

We determined that the conversational AI didn’t understand the customer issues well enough to provide the right solution by assessing the current chatbot feature. To address the problem, Virtusa:

  • Built a natural language processing (NLP) Optimization Framework to improve performance

  • Extracted and fed chat history through the Linguistic Analysis solution to identify similar and different intents. For example: “Why did I get charged?” vs. “Are you going to charge me?”

  • Used the data from the Linguistic Analysis for training to improve the chatbot NLP engine

  • Completed Bot NLP testing and continuous improvements were made in two weeks cycles

  • Created API connectors to backend banking systems to integrate banking functionality
Virtual Agents Solution
The Benefits

Adding context to the chatbot feature helped the bank:

  • Increase chatbot containment rates by over 30% 
  • Lessen live agent call volume by 30% 
  • Decrease the amount of inquiries escalated to live agents
  • Reduce customer service operation costs
  • Improve customer satisfaction
  • Realize ROI through customer interactions


Does your chatbot understand your customers?

Learn more about Virtusa’s Effective Virtual Agents Solution

Related content