In the Business of Customer Happiness - NUMI Pepper x Sparkasse Marburg-Biedenkopf

Sparkasse Marburg-Biedenkopf (SKMB) is the largest independent public credit institution in its business area, the district of Marburg-Biedenkopf, including the university town of Marburg, where it is based.

As a member of Sparkassen-Finanzgruppe, its business activities are guided by the core principles that have defined the mission of Sparkasse for two centuries: to offer as many people as possible from all walks of life opportunities to invest their money, to borrow funds for their endeavors, and to perform their everyday non-cash payment transactions. In 2019, Sparkasse Marburg-Biedenkopf had total assets of 4 billion euros, 817 employees (including one social robot), and 38 branches.

NUMI Pepper x SKMB

In today's digital world, all credit institutions are one click away. Thus an essential element of customer loyalty is eliminated, namely the physical visit to 'their' bank, where customers and bank employees meet face to face and build relationships. In response, credit institutions are increasingly trying to transform the less frequent visits to their offices into an exceptional experience that gives customers good memories and strengthens their relationship with the institution.

Apart from customer behavior, digitalization also has an impact on the internal processes of credit institutions. Automation is advancing, and simple tasks no longer have to be done by humans. This leads to fears among several employees of being left behind in the wake of digitalization, which is not conducive to the acceptance of digital technologies in the workforce.

Another recent development with a considerable impact on the relations between customers, employees, and their bank is the COVID-19 pandemic. With everyone being forced to hide behind facial masks and to keep their distance from one another in order to prevent the virus from spreading, relationship building has become much more difficult.

How do we facilitate relationship building in the age of digitalization and social distancing? To SKMB, it was important that customers, when visiting their SKMB branch, be offered an experience that would sustainably strengthen their relationship with SKMB. Employees should understand that digitalization does not threaten them but rather facilitates and enriches their working lives. And both customers and employees would have to be protected from infection.

Enter NUMI Pepper, a social robot developed by Softbank Robotics and endowed with relationship-oriented communication skills. SKMB envisioned that meeting and interacting with such a robot could elevate visiting and working at SKMB to a special experience while also contributing to human safety in the days of COVID-19.

Furthermore, the robot could help to effectively communicate the benefits of digitalization both internally and externally, increasing the acceptance of digital technologies.

The robot would not be just another machine in service of SKMB, but rather a staff member, the first of a new kind of digital-mechanical employees. By deploying NUMI Pepper, SKMB wanted to achieve more than to attract short-term attention with a waving and game-playing toy. Instead, the goal was to identify value-adding use cases and implement them with suitable applications that are sustainably useful and justify the investment in and deployment of the robot (and its potential colleagues and successors) at SKMB in the long term.

Use Cases

Social Media Performer

SKMB has a strong interest in expanding its presence in social media and is always looking for appropriate instruments to that end. The Instagram account of the robot Luna Pepper of Sparkasse Bremen inspired the idea that SKMB's robot could be a social media performer, an actor, and a model in photos and videos for SKMB's social media outlets that pick up on topics of interest to SKMB and its customers. Having a robot appear in these performances is not only unusual and thus attractive, but it also sends a strong message to viewers about SKMB's commitment to digitalization. The materials to be produced would have to be lucid, concise, and lively, as well as conducive to the desired public image of SKMB.

For this use case, it was decided that the robot would need its own Instagram channel, so one was created in August 2019. This channel is the outlet for all kinds of image and video content related to the robot and SKMB. In the images and videos published there, the robot performs in various settings and roles. For each performance, a dedicated application was developed for the robot.

Experience with the robot's Instagram channel has shown that with relatively little development effort, positive effects can be achieved. Content continues to be published regularly on this channel as well as on Facebook and YouTube, and SKMB is confident that the robot's increasing social-media popularity will promote its overall image and the image of SKMB.

Customer Trainer

There are many situations in SKMB's day-to-day business where customers need to know how something works or what benefits a product or tool offers. The necessary explanations are often routine, so SKMB decided to use the robot for training customers in these cases. The idea was that the robot would be able to explain standardized functions and processes to each customer in a constantly friendly, patient, and competent way, using animated spoken commentary complemented by presentations on its tablet screen.

In this video, the robot explains a technical term ("Sparkasse"). This video is part of a series of educational videos that were produced to train customers in understanding important terminology in the context of SKMB's activities. Each video’s length is up to one minute, which requires the explanations to be concise and to the point.

For another training app, existing short videos and other materials were integrated with the robot's spoken commentary to develop performances that provide information about the opening hours of the central SKMB branch in Marburg; assistance for filling in online and paper transfer forms; and information about advantages and configuration of the Sparkassen app. In addition, the robot can offer a step-by-step guide to setting up the Sparkassen app.

Presenter

In 2019, SKMB opened a brand new branch in the city of Biedenkopf, Hesse. The building is state of the art in terms of both architecture and digital technology. Who could present this achievement of SKMB's to a wider audience in a video to be published on the Internet? SKMB's digital-mechanical employee, the social robot NUMI Pepper, seemed like the obvious choice.

With their versatile, communicative capabilities, social robots like NUMI Pepper are well suited to act as presenters. Depending on the requirements of the presentation, the robot's performance may integrate speech, dialogue, gestures, animations, and movement. The robot's tablet screen may display audiovisual content that supports the presentation.

To present SKMB's new Biedenkopf branch, a complex presentation app was developed for the robot that consisted of multiple scenes, each highlighting a particular feature of the new branch. The scenes could be performed independently of one another by selecting the current scene from the robot's tablet screen. This facilitated filming the robot's performance on location considerably. Depending on the subject and goal of a scene, the scenes were either solo performances by the robot or involved NUMI interacting with people working at the Biedenkopf branch. For each scene, a script was developed that detailed what the robot and the human participants would have to say and do. For the robot, the script instructions were translated into animations, locomotion, and dialogue rules that were implemented in the presentation app. Special care was taken to make each performance lively, interesting, and appropriately humorous.

Knowledgeable Conversationalist

Working with its customers is a daily, communication-intensive task for SKMB that traditionally has fallen to its customer advisors. However, many communication requests are about routine concerns, whose repetitive nature can quickly irritate and tire humans, whereas machines are not bothered at all. In addition, machines can be available around the clock for inquiries from customers without long waiting times. Therefore, SKMB has integrated automated conversation solutions into its customer communication services, including both natural language dialogue systems over the telephone and text-based chatbots embedded in its website.

The design of social robots makes them ideal for communicating with people. Unlike chatbots or telephone-based systems, they are part of people's immediate surroundings and can therefore communicate with them face to face. That is why SKMB saw the opportunity to use its social robot NUMI Pepper for communication with customers. However, there was a problem: NUMI lacked the competence to engage customers in dialogue and to hold its own as an informed and helpful conversation partner. Obviously, this requires substantial knowledge about the diverse range of topics at SKMB. Developing the necessary dialogue competence in-house was not feasible, so SKMB looked for external solutions and found one in Linda, a chatbot developed by Sparkassen Finanzportal (SFP). Linda is already working as a virtual assistant at SKMB's service center, where she answers customers' routine questions outside business hours.

The basic idea for the NUMI-Linda integration was for the robot to act as the point of contact with the customer. Any questions it would be unable to answer from its own knowledge, it would send to an instance of the chatbot Linda over the Internet and then perform Linda's answer for the customer. The connection to Linda was established via WebSocket, and data was exchanged in JSON-format. SFP provided access to a web-based chatbot editor that allows SKMB to customize Linda's answers for spoken delivery by the robot. This means essentially reducing the length and complexity of the answers. The app developed for the robot was named "MultiChat" because it has the ability to accommodate multiple chatbots, both local and remote.

Exhibition Guide

SKMB regularly hosts exhibitions of artists and craftspeople in its facilities that customers can visit during opening hours. However, SKMB cannot provide knowledgeable guides for these exhibitions, as the expertise of its advisors lies elsewhere. But NUMI Pepper is available as a programmable exhibition guide. In this role, NUMI moves from one exhibit to the next and provides a detailed verbal explanation of each:

NUMI needs to be able to locate each exhibit, move towards it, and provide the relevant explanation. The exhibition guide app uses so-called ARUCO markers for this purpose (the red-and-white square patterns mounted on posts in the video). Each exhibit has its own marker that both serves as a landmark that the robot can move towards and provides a visual code that the robot can scan to identify the exhibit and retrieve its corresponding explanation from its internal database. The robot first locates all ARUCO markers in its environment so it can find them later. Then it waits for human visitors. In guided-tour mode, it presents all exhibits in sequence, navigating to the frame of each located marker. A special home marker enables it to return to its original position when the guided tour is done.

Receptionist

Customers entering a branch or department of SKMB should be received properly to positively shape their perception right from the beginning of their visit. SKMB needed a receptionist who is friendly and competent and gives visitors a good first impression. Because of its capabilities, the robot appeared suitable for this role. Positioned in the entrance area of a branch or department, the robot perceives customers as they enter, approaches and greets them in a friendly manner. If the customer has an appointment with an advisor, the robot sends a notification that the customer has arrived. While the customer is waiting for their advisor, the robot offers to help them pass the time by showing them information about SKMB's products and services. For this use case, a proof-of-concept app was developed:

For identification, the customer has a QR code scanned by the robot's tablet camera. The code contains the ID of an appointment record in the robot's database (in the future, this ID will point to the relevant entry in SKMB's appointment management system). In turn, the appointment record identifies the advisor and sends them an email stating that their appointed customer has arrived. When the robot has sent the email, it informs the customer that their advisor will pick them up shortly and show them information about KWITT (a service for sending money over the phone) while waiting. The customer can, of course, decline this offer.

If the customer has forgotten their QR code, the robot asks for the advisor's name and retrieves their email address on that basis. In still another mode, the QR code contains the advisor’s email address, which is then used directly to send the notification.