Business challenge

Mizuho Bank, Ltd. sought to drive highly effective customer interactions by injecting intelligence and voice recognition capability directly into the conversation to dynamically guide the flow.


The customer conversation can be analyzed in real-time using cloud-based natural language processing (NLP) algorithms to instantly formulate the best question path for the agent to follow.



in time required to respond to customer contact center inquiries

customer service level

by delivering a more satisfying and personalized experience

quality and volume

of training for call center operators by using real-time content analytics

Business challenge story

Intelligence needed for customer interactions

Mizuho Bank needed to improve the effectiveness of its contact center agents. The key underlying challenge was the complexity of the information agents needed to navigate to guide customers through an inbound call session. Although the most veteran and experienced agents had the familiarity necessary to quickly “read” customer needs from the exchange, the majority of agents—and especially new trainees—didn’t.

At the core, the bank wanted to bolster agent performance by adding a layer of automation and intelligence that would perform the logical associations that agents were required to make effectively during the call. Instead of relying on agent experience, which can vary considerably, the bank wanted to identify and analyze contextual clues during the course of the conversation and use the insights to guide agent discussions in real time.

By giving our agents real-time insights into each customer’s needs, we’re creating the basis for a more satisfying customer experience.

Tetsuhiko Saito, Chief Marketing Officer, Mizuho Bank Ltd.

Transformation story

Real-time customer conversation analysis

Mizuho Bank engaged IBM Research to develop a First-of-a-Kind (FOAK) customer contact optimization solution that employs advanced speech recognition technology and IBM® Watson™ content analytics software and runs on the SoftLayer® cloud services infrastructure. After converting the customer’s voice to textual data, the solution applies natural language processing (NLP) algorithms to each interaction to infer the customer’s specific needs or goals at each point of the conversation. Correlation algorithms run against historical customer service records and then formulate the optimal response, which is delivered in real time as a prompt on the agent’s screen. Based on the correlation’s measured accuracy, the solution continually “self-teaches” by adapting the algorithms.


Results story

Duration of customer interactions reduced

The solution reduces the average duration of customer interactions by more than 6 percent by enabling contact center agents to more efficiently sense and respond to customer needs. By delivering a more satisfying and personalized experience, the solution increases customer retention.

The solution also significantly reduces contact center training requirements by using real-time content analytics to augment employee knowledge.

Mizuho Bank Ltd.

Mizuho Bank Ltd. blog

Based in Tokyo, Mizuho Bank Ltd. is the integrated retail and corporate banking unit of Mizuho Financial Group, one of the largest financial services company in Japan. With more than 500 branches, Mizuho Bank is the only bank to have branches in every prefecture in Japan.

Solution components

  • Banking: Next Generation Experiences
  • Content Analytics with Enterprise Search
  • IBM Cloud Managed Services

Take the next step

To learn more about Watson, please contact your IBM representative or IBM Business Partner, or visit the following website:


View more client stories or learn more about IBM Watson