Revolutionize non-face-to-face channels with AI

Call center digitalization and future contact center vision



Due to the impact of COVID-19, financial institutions have been restricted from operating at branch counters, which used to be the first point of contact with customers. Now the call center plays an important role. Japan Post Bank Co., Ltd. is working on digitalizing call center operations intending to reform customer contact points for the future. Takayuki Tanaka, Managing Executive Officer, has high expectations. He says, "We would like to actively utilize various mechanisms such as AI systems to create a call center that is friendly, polite, convenient, and easy to use as a contact point for customer feedback."
We spoke with Minoru Yamaura, the director of the call center department at Japan Post Bank, and Yuichi Aoyagi of NTT Data, who supports financial institution clients with the power of digital technology.

Customer's issues

  • At the call center, the agents who are at the forefront of customer service play main roles. In order to hire and grow "agents" and retain them at call centers for years to come, it is necessary to devise ways to create a workplace environment where people can work comfortably.
  • When face-to-face becomes difficult with COVID-19 this time, the number of calls at the call center, which is part of the non-face-to-face channel, is about twice as many as usual, and it is necessary to maintain the service with the minimum number of personnel.


  • We aim to achieve CX (Customer Experience) improvement and workload reduction by promoting call center digitalization and contact center transformation.
  • By introducing an AI system to the call center, workload for various tasks is expected to be reduced while maintaining or even improving customer experience. For example, the average handle time (AHT) for new agents was reduced by about 30% compared to the case without AI. In addition, by monitoring agents and customers in real time, the AI system will improve the quality of service.
  • "Chatbots (robots)" will be able to respond to inquiries about products and services from customers 24 hours a day, 365 days a year.
  • As a BCP response, we are considering having some agents work from satellite call center (or at-home call center), which will lead to work style reforms and ES (Employee Satisfaction) improvements

The key point is "people"

-- First, please tell us about the position and role of the call center in Japan Post Bank.

Mr. Yamaura
Our customer touch points are broadly divided into "face-to-face channels" and "non-face-to-face channels," but the main customer touch points are face-to-face at stores. The call center will be positioned as the most human-touch customer contact point among non-face-to-face channels. When it became difficult to meet face-to-face due to the COVID-19 this time, the call center, which plays a part in the non-face-to-face channel, showed its presence. In April and May 2020, number of call center calls reached about twice the normal number. We believe that the existing system to provide service centered on face-to-face channels will need to be reviewed in the future.

The call center has two roles. One is to firmly "respond to customer phone inquiries", and the other is to "share information within the company" with voice of the customer.

For the first role, while our products and services are being digitalized (Yucho Pay app, Yucho Authentication app, Yucho Bankbook app, etc.), it will be critical for our call center to navigate through our customers who are not accustomed to the Internet environment and smartphones, including elderlies.

For the second role, it is important to be able to support the realization of customer-oriented business operations by linking customer feedback within the company through telephone and running the "Improvement PDCA Cycle" of our products, services, and administrative procedures.

Diagram: Positioning and role of call center for Japan Post Bank

Diagram: Positioning and role of call center for Japan Post Bank

-- You said that you were the director of the Sapporo Call Center for three years from 2017. What were the problems and challenges you actually experienced there?

Mr. Yamaura
90% of the work of call center directors is about "people." How to hire, grow, and work for a long time. At the call center, the agents who are at the forefront of customer service play main roles. How we create and maintain an "inverted pyramid of support" where agents are supported by assistant supervisors (ASV), and they are together supported by supervisors (SV) is critical in having agents work happily. This leads to call center's Operational Excellence (Competitive Advantage).

When I was the director of the Sapporo Call Center, I had to face lots of challenges. Sapporo is a city where call centers are concentrated so that it is called "Call Center Ginza." We made efforts so agents will "choose our call center," "stay at our call center for a long time," and "work at our call center happily." We had our candidates all visit our call centers. For training we combined classroom lectures with taking actual calls so agents can learn the skills more naturally. For evaluation, we developed a system that rewards people who have worked hard.

Diagram: "Inverted pyramid of support" and call center business overview of supervisors

Diagram: Inverted pyramid of support and call center business overview

Digital brings reduction of workload and improvement of response quality

-- Under such circumstances, you are introducing AI system to the call center this time. What types of effect do you expect from the use of AI?

Mr. Yamaura
When we categorize the skills required for agents, they can be divided into three groups: "consciousness," "skills," and "knowledge." Among these, "consciousness," that is, feelings of valuing customers and hospitality, is a part that only people can do, and AI cannot solve it. Also, in "skills", the part of thinking ability such as logical thinking and imagination necessary for dealing with customers are the areas that require human power. On the other hand, of "skills", input skills (seeing), output skills (writing), and "knowledge" are where AI can show its strength.

Diagram: Skills required for agents and "areas where AI can support"

Diagram: Skills required for agents and areas where AI can support

Mr. Yamaura
Based on this premise, I expect five results from the introduction of this AI. The first is to reduce agents' workload of creating the call log. Agents usually leave the summary of the conversation with the customer, but AI can convert speech-to-text, eliminating the need for agents to create conversation summary. The second is to reduce agents' burden of looking up the relevant knowledge. AI figures out what the call is about and automatically suggest a list of relevant knowledge that the agents will need during the call, saving agents time and efforts to look for the right knowledge. This is especially helpful for new agents. We currently provide pretty extensive new agent training, but we may be able to reduce the training period if we can reduce agents' workload of looking up the right knowledge.

The third is to reduce the false response rate. Since AI automatically suggests the necessary knowledge, agents develop a habit of always referring to the knowledge when responding to the customer, and this is expected to reduce false response. The laws and systems of financial knowledge often change, and agents must update their knowledge. With the automatic suggestion of knowledge by AI, agents can always check the latest knowledge. The fourth is to reduce supervisors' burden of monitoring work. Supervisors monitor the exchanges between customers and agents by listening to the audio for quality check, but if the audio is converted into text, supervisors can accurately grasp the exchanges and provide timely support. And the fifth one is to promote hiring and prevent employee churn. We believe that the introduction of AI will reduce the burden of various tasks and create a workplace environment where people can work comfortably.

Diagram: Overview of AI system introduction to call centers and expected effects

Diagram: Overview of AI system introduction to call centers and expected effects

Proof of Concept that gave us conviction

-- Could you tell us about the timeline and process of introducing the AI system this time?

Mr. Yamaura
We consulted with NTT DATA in 2015 when I was working on BPR as the manager in charge of the Operation planning department. Originally, I was thinking of introducing it to a call center for the JP group, which handles telephone inquiries from the post office called the partner center.

Mr. Aoyagi
Since that time, we have been thinking that AI can contribute to the efficiency of Japan Post Bank's business from the perspective of digital labor, and we were proposing AI solutions suitable for Japan Post Bank's business.

Mr. Yamaura
After that, when creating the FinTech vision as the manager in charge of the system management department (and also the corporate planning department / FinTech PT), I paid attention to the importance of AI and knew that if we introduce it to the customer call center we could improve the quality of service and work efficiency. Eventually, in 2018, when I was the director of the Sapporo Call Center, a decision was made to introduce NTT DATA's AI system, and a Proof of Concept (POC) was conducted from October to December of the same year.

Mr. Aoyagi
NTT group owns AI technologies called "corevo." One of the characteristics of "corevo" is its high level of speech recognition ability, which has been refined by NTT Laboratories. By utilizing "ForeSight Voice Mining®" of NTT TechnoCross Corporation, which is an AI system using "corevo", we proposed that the comprehensive strength of the NTT Group could solve the problems faced by Japan Post Bank, and we were selected.
"ForeSight Voice Mining®" utilizes the voice recognition of NTT Group's AI technology "corevo®", emotion analysis technology to convert the calls with customers into text in real time.
It is possible to support agent operations by automatically searching for related FAQs from the text and automatically creating summaries. This time, the important point for introducing our solution to Japan Post Bank was to increase the accuracy so AI can support humans more precisely. This included brushing up the speech recognition technology and tuning it so that appropriate answer candidates and materials can be presented.

-- Did you experience any difficulties or have anything you specially devised during the PoC?

Mr. Yamaura
In order to improve the FAQ correct answer rate of the AI system, we analyzed the inquiries / answers from customers and the feedback results of agents, created FAQs focusing on frequently asked inquiries, and integrated similar / duplicate FAQs as much as possible. As a result, we were able to improve the correct answer rate to about 80% at the end.

Mr. Aoyagi
After the successful PoC, NTT DATA and NTT-TX have proceeded with system construction, and commercial deployment of ForeSight Voice Mining has started in January 2021. Call centers are made up of a combination of different systems. When introducing an AI system, we have to link data with existing systems (CTI, CRM, etc.), which is challenging. In that respect, NTT DATA's long-standing track record of providing stable financial IT services that require high reliability to Japan Post Bank paid off. We are working every day to provide the system in the optimum condition, taking advantage of our system integration skills.

< Comment from Mr. Tetsuro Ogushi, Chief of Fukuoka Call Center, actual user of the system in the PoC >

My staff who participated in the PoC told me that some of the PoC participants were new agents with little experience, but by having the AI system present answer candidates based on conversations with customers, they were able to quickly search for knowledge, have smooth conversation, and get used to the call center routine quickly. From the quantitative point of view, we saw the most obvious effect of shortening the average handle time (AHT) with new agents. They were able to shorten the AHT by close to 30% compared to the case without using AI. In addition, supervisors were able to accurately grasp the content of conversation with customers, and they were able to provide better support.

People at the headquarters call center department in charge of AI introduction project
(Center) Mr. Takayuki Tanaka, Managing Executive Officer
(Right most) Call center department・Mr. Minoru Yamaura, Director
(2nd from the right) Call center department・Mr. Kazuki Oishi, Group leader
(Left most) Call center department・Mr. Kohei Kumagai, Manager
(2nd from the left) Call center department・Ms. Yumie Kawano, Chief

Aiming to turn it into a contact center

-- Lastly, could you talk about the future plan for Japan Post Bank's call center?

Mr. Yamaura
The AI-based FAQ search engine "AI-FAQ" is a collection of frequently asked questions from customers and answers to them, so we can call it our "organizational knowledge."
We will continue to update it in the future.
Although we cannot do it overnight, applying this organizational knowledge, our ultimate goal is for us to be able to respond to inquiries from customers regarding our products and services with a "chatbot (robot)" 24 hours a day, 365 days a year. If the chatbot cannot solve the problem, we will support " human support chat" to enhance the chatbot's presence. If the chatbot cannot handle it, we will deal with it by "telephone" and at a "physical branch."
Our future plan is to transform the conventional call center that only supports telephone into a "contact center (chatbot + human support chat + telephone)".
As a BCP response, we will consider having some agents work from satellite call center (or at-home call center). During COVID-19 (April - May 2020), we reduced the business hours of the call center to the core time (9:00 to 17:00), which had the most number of incoming calls, and we implemented split operations with the minimum number of personnel required. We are taking infection prevention measures through efforts like this, which secures the distance between agent seats. As an infection prevention measure in preparation for the 2nd and 3rd waves of the COVID-19, we have installed acrylic board partitions between the agent seats (front and side), but in order to secure a more appropriate distance between the seats we would like to establish a system that allows some agents to work from satellite call center (or at-home call center) and take calls. I think that making satellite call center (or at-home call center) can lead to work style reforms and ES (Employee Satisfaction) improvements for agents, assistant supervisors, and supervisors. I have great expectation in NTT DATA's ability to propose and execute digital innovation in order to realize the future contact center vision.

Diagram: Future contact center vision (image)

Diagram: Future contact center vision (image)

Director Call Center Department Marketing Division
Mr. Minoru Yamaura

1992 Graduated from Waseda University Graduate School of Science and Engineering, Department of Mechanical Engineering (Management Engineering)
Joined Mitsubishi Research Institute in the same year
Engaged in consulting business centered on automobile / precision equipment manufacturers, electric power / gas companies, and JP Group companies
2008 Seconded to Japan Post Bank / Operation Planning Department as manager in charge, then transferred
2016 System Management Department as manager in charge (and Corporate Planning Department FinTech Project Team)
2017 Director of Sapporo Call Center, Marketing Department III
2020 Director Call Center Department

NTT DATA Corporation
Senior Manager Postal Services & Social Finance Division First Financial Sector
Mr. Yuichi Aoyagi

2000 Graduated from Kyushu University, Faculty of Letters, Department of Human Sciences.
Joined NTT DATA in the same year.
Engaged in bank accounting system development and new service planning for banks.
Financial Business Promotion Department Group Business Strategy and Open Innovation Business Emergence Office, Engaged in financial business strategy formulation and open innovation activities for financial institutions.
2018 Senior Manager Postal Services & Social Finance Division First Financial Sector
Engaged in new business proposals for Japan Post group.


Due to the privatization of the Japanese postal service, the company was established as Yucho Co., Ltd. (privatization preparation company) on September 1,2006.
The Japan Post Group was established on October 1,2007, and the business started under a new trade name, Japan Post Bank Co., Ltd. The company was listed on the First Section of the Tokyo Stock Exchange on November 4,2015.
Under the management philosophy of "Aiming to be the most well-known and trusted bank with the voice of customers as a compass for the future, we will innovate retail business into a new form by realizing complementarity between the physical and the digital. As a result of these moves, we will provide "reliable and thorough" financial services "safely and securely" to anyone and everyone throughout Japan.