Management Engineering - Leadership & Innovation

Project Example

Projects

1.1 Introducing Banca Mediolanum 1.2 The Current Meaning 2.1 Technological Analysis 2.2 Social Analisys 3.1 Giving a New Meaning 3.2 The Manifesto Executive Summary 4.1 Introduction 4.2 Customer Segments 4.3 Problems & Needs 4.4 The Solution 4.5 Channels 4.6 Unique Value Proposition 4.7 Unfair Advantage 4.8 Metrics 4.9 Economical Feasibility Analisys 4.10 In a Nutshell: Final Considerations 5.1 Meet the Team 5.2 Team Development What is happening nowadays is that Mediolanum wastes time and money in a call center service that cannot always meet customers’ needs. How to solve it? By changing the meaning leveraging on Artificial Intelligence. Why should people call when they can connect? Why should customers repair when they can build? To answer these questions, it has been necessary to analyze Banca Mediolanum problems, digging also in social and technological trends. This pursuit made possible to be “Costruito intorno a Te” no more only a motto, but a way of thinking that fits on people’s lives. The bank and the people face innovation together in a unique path that aims at a complete personalized account through an Omnichannel strategy. The goal is the maximization of customers’ benefits without losing Mediolanum’s perspective. It takes time to be strong: a path with safe camps is needed to reach the peak. Thus, the Artificial Intelligence learns through data connecting with people to build the uniqueness. The final purpose of this idea is to create tailored accounts that ease people’s lives, save time and make more accessible the communication with their bank. ( Financial year 2017, for Mediolanum Group ) Setting in motion the issue presented in this project is the Customer Service Department at Banca Mediolanum. Being a Bank with no brick and mortar locations, interaction with the customer happens through various channels:  Website  App  Phone  Smart TV  ATM Network  Family Banker The core of the problem is the Phone-based Customer Service. Different from all other ways to interact with the bank, the customer service requires employees to function and this creates a variety of different consequences: Mediolanum managed to create a very efficient call center, with low response time and well-trained employees, so where does the problem lie? The problem with Mediolanum’s Call Center is that it is a Call Center. This is a very high performing call center, but it is still a call center in the traditional sense, and this encompasses all its typical characteristics:  Labor Intensity  Relying solely on people to solve unexpected situations  Focus on problem solving  Communication is initiated by the customer The freedom for customer to interact with the bank wherever, whenever and however they want. Therefore, the bank has to offer a fast and efficient phone customer service. There are over 450 employees working in the Call Center: they represent over 1/5 of all Mediolanum’s workforce. Th e call center is only resolving the 5% of all requests made to Mediolanum. Moreover, they are mostly repetitive tasks. The company has and still is investing in innovative and Digital communication channels and has decided to heavily invest in the call center to align it in performance with other channels. What is causing the problem is the traditional idea of call center: it simply cannot compete with the speed, capacity, availability and convenience of digital channels. It’s possible to encapsulate the meaning of Mediolanum’s customer service in a very simple, yet clear way: The analysis of the environment and the trends for the future is made on two main categories:  Technology: as an enabler for the innovative solutions that the call center is looking for. “You can spend a lifetime building the fastest train ever. Or you can simply get a plane.” \ The reason a customer calls is most of the time a problem that he hasn’t been able to solve by himself or with other channels made available by the bank. The interaction has the purpose of “repairing” the relationship between the customer and the bank, which has been jeopardized by a problem. Communication is initiated by the customer. He is the one asking for information and this translates surely in a two-way communication, but with the customer as the initiator and with the employee only as a provider of answers.  Social Environment : as the challenge is tied to the interaction with the human. Emerging trends and key concepts in those areas are shown below and the most important ones will be analyzed in detail. Automation can be defined as “the technology by which a process or a procedure is performed without human assistance”. This phenomenon is going to increase, due to two things: robots are becoming more precise and efficient than humans in the execution of repetitive tasks. Also, the manpower that could be moved from low-level skill positions to higher-level ones would lead to greater value for the company. In connection with this point, the repetitiveness of a job can lead a human being to depression, as Dr. Greg Couser explained with his studies about repetitive tasks. According to this idea, automation could solve the feeling of alienation, substituting human beings in these kind of jobs. Artificial Intelligence is a computer science that tries to imitate the way in which the brain works, analyzing a huge quantity of heterogeneous data in the process. Nowadays, the use of AI technology has rapidly diffused through various products, services and industries . However, 34% of the participants claimed that they have never interacted with artificial intelligence technology whereas 84% of them is using devices/services based on AI without knowing it (Source: Accenture). This result proves that most consumers tends to misidentify AI. Teradata interviewed 260 IT and business decision-makers and found out that 80% of these organizations already have some form of AI in their production. Additionally, 30% percent of enterprises are planning to expand their AI investments over the next 36 Big Data Automation Ecosystems Artificial Intelligence + Machine Learning Data Overload Blockchain 5G AR & VR Internet access in developing countries months and 62% expect to hire a chief AI officer in the future. These results prove the necessity of AI presence in businesses and reveal the potential of technology-push innovations. Differently from the past, the product is no more at the center of the purchasing process: the customer is. Clients want a unique and personalized product and an exclusive service that makes the customer feel more important and part of the company. This aspect is very important because today there are many competitors and so customer care becomes the crucial point on which concentrate the activity of marketing and development. Another key aspect is customer satisfaction: it must be measured and monitored continuously by the company, taking in consideration that it is a momentary status and it varies enormously among individuals. Companies face a new problem: consumers are increasingly attentive to quality and less willing to tolerate mistakes. Firms must address buyers’ impatience, listen to his needs, communicate according to the brand identity and in a coherent way. One way to improve the customer experience is through an omnichannel strategy, in which all the channels are connected and integrated. This continuous connection needs to offer the same level of service in each channel, giving the idea that they are always interconnected with the company. From a business perspective, the key aspect is the integration between the firm’s vision and the communication strategy. This allows to offer an optimal customer experience providing a long lasting customer journey and to avoid repetitive information, incoherent communication and unsatisfied customers. Impatience Low Attention Span Fear of Technology Customer Experience Omnichannel Approach Advertising Overload Social Media Falling trust in companies Privacy Concern Considering that nowadays technology is embedded in human life, an huge part of the population is affected by Technophobia . This fear regards frequent changes in technology, since people cannot keep up with this phenomenon falling in the risk of losing contact with reality. In these days, a new trend is emerging: the fear of Artificial Intelligence. There are two opposite ways of thinking about this topic. On one side there’s a pessimistic perception that believes that an extreme development of A.I. will be fatal for the human kind. On the other side, there is an optimistic perception that highlights the benefits of it, due to the positive impact on production thanks to automatized systems. Customers start perceiving A.I. no more as something that can hurt society, but rather as a help. Customer are more open to technological innovation when it comes to save their time or money. On the other hand, developers have to ensure human health and enhance people’s quality of life to be accepted in society and to solve the fear of technology. Looking at the Social and Technological environment analyzed in Chapter 2, the call center problem can be analyzed with a new perspective. After having seen the state of technology and of social interactions, today’s meaning “Call to Repair” feels dated already. The analysis shows that new possibilities are emerging and companies around the world are embracing the change. It is possible to see that the relationship with the customer has changed, becoming more intimate, personalized and informal. Nowadays, firms want to have a very personal and friendly communication with customers, making them active actors. Analyzing the development in technology, it has emerged that leaps in A.I. and Machine Learning have given the means to move in two directions: computers think and interact more human-like and at the same time that are able to integrate data from various sources, learn patterns and get better exponentially as time passes like no human can do. According to this, each time that the customer interacts with the company, the latter can learn and get to know the client better, offering each time a slightly better service. The sum of these small improvements creates a huge benefit in the long-run for both parties, improving satisfaction and loyalty. Using technology as an enabler and having a customer that is an active part of the interaction it’s possible to give a new meaning to Mediolanum’s call center: These two words are causally connected: a connection with the customer is needed to build a better experience and a better service. The new vision and meaning for the change is encompassed in the manifesto. It sets the guidelines for the development of the solution and shapes the pillars on which the future of Mediolanum’s customer service will be based. Ffff Helping the employee in achieving i ts full potential, eliminating repetitive work. Giving the customer tools to get the most out of his banking experience. Through an omnichannel and a proactive approach, the customer and the bank can connect with each other exchanging information and creating a jointed value. Creating value for the customer, so that he will be loyal to the bank, and the bank will be loyal and respectful to him. It takes time to be strong, the relationship and the value have to be built overtime . Once that the vision for the future has been defined, the next step is shaping it into a detailed and feasible plan. This has been done considering the Social and Technological situation detailed in Chapter 2. The solution to the problem will be presented using the Lean Canvas, also to give an idea of the process that has been followed. Understanding who is the customer and which problems can emerge. Presenting the Solution in detail, outlining the interaction with the customers and the various touchpoints involved. Understanding the unique advantages that the whole company can have from the implementation of the solution. Defining the performance management and the economical feasibility of the project. Customers have been segmented relying on Mediolanum’s four bank accounts types. For this analysis, each segment represents the different users’ clusters: starting from these prototypes of personas each segment has been developed. Works as a Shop Assistant, Mid-Low Income. Loves traveling. Embraces new technologies and change. Wants a simple and convenient banking solution. Teresa, 22 Owner of his own Advertising Agency. Travels for business at least 6 times a month. High income invests in the Stock Market. Wants a bank with custom solutions and that is always available. Roberto, 43 Worked for 40 years as a Mailman. Lives on his pension. Owns an old Flip Phone. Wants a bank that’s easy to contact that manages with care his savings. Luigi, 76 Working mom & dad. Medium income, paying off house. They see technology as a tool. They want a bank that is reliable and stable. The Bianchis 37, 35 & 7 Expects outstanding service, convenience and experience. Wants things get done quickly and effectively. Need for simplicity. Sees technology as a challenge and as a problem. Need for simplicity. Different generations have different relationships with technology and its linkage to privacy and security. Younger people are not worried about the possibility to lose control over their privacy and data. On the other hand, elders are more sensitive to this issue. This doesn’t mean that they refuse technology, but they use it in a more responsible way, paying more attention to their data. This is one of the problems to be faced is customer data management. A survey shows that 87% of consumers believes it’s important for them to be able to review and manage their online data. However, three fourth of them do not find easy to do it. Furthermore, 90% percent of respondents says security of online financial transactions is crucially important for them, whereas only 30% of consumers trust in banks about their personal data (source: Accenture). It is crucial for banks to inspire confidence and loyalty to meet customers’ expectations. The graph on the right, introduced by Masahiro Mori, is useful to explain the relationship between a human being and an android entity. In 1970, roboticists were driven by the idea that the higher the human likeness, the higher the familiarity between human and robot. Mori affirms that the previous belief is not completely true, since there’s only an apparently biunivocal relation between x-axis and y-axis. According to his theory, there’s a point in which the familiarity drops and the Uncanny Valley is born. Basically, it represents the uneasy sensation people feel in viewing a humanoid entity that looks closely to a person but that it is not human. Experiments to clarify this issue were conducted in the University of California, San Diego, where videos of a human, a humanoid and a metallic robot acting the same movements were shown. The main result is that the human brain cannot elaborate the mismatch between the android’s human likeness and its robotic movements. This incompatibility is considered the cause for the development of the Uncanny Valley. The inability of the brain to completely understand the incongruity between what appears and what it really is, lead to the perception of a “zombie”, or something frightening and not familiar at all, due to a cognitive dissonance between what is seen and what is perceived. Human Likeliness Familiarity Artificial Intelligence has been known by the public for years now, but most of the users today have a vision of A.I. that’s distorted by pop culture. Sci-Fi Movies and TV shows have depicted Artificial Intelligence in a very negative way, or they showed applications that are completely different than what A.I. actually does. Over the last couple of years, the more tech-savvy really started to understand A.I. and its actual capabilities, but still most users don’t know what benefits it could bring. On the contrary, there’s still a percentage of the population that has a negative view of A.I., seeing it as a threat. The solution will rely on the new meaning “Connect to Build” to create something bold and unique for the company. According to this, the purpose of this project is to build a Bespoke Account for each customer, creating a deep connection with him. The implementation will not be a moment but a process that relies on data collected via an omnichannel approach. The solution will be implemented in a Three Step approach in order to transition to the new paradigm more gradually and using the data collected from each step to improve the implementation of the following one. “Change, like healing, takes time” Veronica Roth A detailed description of the implementation, step by step, is presented below. The first step is mainly focused on A.I. learning: each call made to the call center by customers will be recorded and analyzed using a Voice Recognition Engine. In this phase the A.I. system will also learn with the help of the employees: at the end of the call, in case it doesn’t understand the inquiry completely, it will ask the employee what the right meaning was through multiple choice questions. This will allow Artificial Intelligence to improve at a much faster pace, without overburdening the employee. The A.I. is helping the employee to answer quickly and more effectively the requests made. A.I. will be integrated in the computer screen of the operator and it will:  Show to the operator the location of the call.  Through the Voice Recognition software try to provide a guess for the answer that the client is seeking, so that the operator will not need to look for it.  Provide contextual information for the call and the user. The customer will not interact directly with A.I. in this phase, as it still isn’t trained enough to withstand a full conversation. The number of employees remain untouched, but now the operator can solve inquiries faster and more effectively. AI requesting info on the call to improve. AI offering contextual information to operator. The client is informed about the future introduction of A.I. through all Mediolanum channels. It’s crucial to communicate the integration plan in the right way, highlighting the benefits that A.I. can bring and reassuring customers about the reliability of the system. The first interactions between customers and the A.I. can give critical feedback about customers’ behavior. For this reason, a Pilot Program will be started, offering customers the A.I. Integration described in Step 2. This pilot program will be tested on clients that are profiled as “Innovators”: more inclined to innovation and technology (estimated to be 2.5% of Mediolanum’s customer base using the product adoption curve). Starting from the 1.044.000 Mediolanum customers, it’s known that the average customer makes 19 interactions with his bank per month across all channels (source: Accenture). Since the 5% of all requests are handled by the Call Center, this results in 297.540 calls/year made by Pilot Program participants in the first year: it is a big source of raw data for the A.I. to start perfecting the actual interaction with customer. The A.I. integration will require a specialized team of engineers and a manager that will oversee the whole system. This will result in a 5 people A.I. Specialist team + 1 Manager. The first phase is expected to last between 9 and 12 months. The artificial intelligence engine is continuously learning, but now this is not the focus anymore. The amount of sample data gathered in the First Step is enough to ensure a sufficient level of reliability and knowledge of the possible scenarios. The Second Step represents the main turning point regarding A.I. integration. From now on, all incoming calls will be answered by the A.I. This phase is particularly crucial because here the solution needs to tackle the Uncanny Valley problem. As described in Chapter 4.3, the Uncanny Valley problem emerges every time an A.I. system is implemented. In the last decades more and more companies have been trying to overcome this issue by moving right on the chart and climb up the curve. In this implementation a different approach has been chosen, and it can be summarized as: In this context there is no need for hyper realistic simulation of human behavior, as customers are just looking to get things done efficiently. A good example is what Google has done with its Assistant. The Artificial Intelligence is used at its full potential in the Speech Recognition and Backend information retrieval, but the communication with the customer is kept still human-like only to a certain degree. The angle taken by the proposed solution and Google Assistant is the same: usefulness over human-likeliness. The Artificial Intelligence voice assistant will have a female voice and it will be called M.I.A. (Mediolanum Intelligenza Artificiale). Generally, Artificial Intelligence works by comparing the real-life data retrieved (in our case the request made by the customer) with a set of known situations or combinations of them. The output is the recognition of one of the known situations with a certain degree of confidence. Human Likeliness Familiarity “People don’t want a system that’s human, people want a system that’s useful.” A.I. 93% Scan the QR code to hear a Live Demo on YouTube™ of Mediolanum Intelligenza Artificiale in various call center requests scenarios. Requests are analyzed and divided in two categories:  Standard Requests (confidence over 90%) This kind of requests are automatically handled by A.I., with no need for human intervention.  Specific Requests (confidence under 90%) This kind of requests are transferred to an operator automatically by the system. The A.I. will inform the customer that the call will be transferred to an operator, who will immediately have on his screen the whole history of the conversation with the A.I. and other useful data. Since the system is continuously learning, the percentage of Specific Requests will decrease overtime and more complex questions will be answered automatically. The customer will still be able to speak with an operator at any time. In that case the call will be handled as if it was a Specific Request. The A.I. is now well trained and will be shared as backend by all other channels:  App  Website  Siri & Google Assistant Details on the integrations and mockups of design and interaction can be found in the “Channels” Chapter 4.5. A Proactive Approach is an important element in the meaning of connection. The A.I. is trained and directly connected with people, knowing most of their needs and behaviors thanks to the interconnection of the omnichannel approach with all the different devices. Taking all this data together, the A.I.’s algorithm is able to give advice that could create economic and financial value for the customer, such as money and time savings. This advice service will work through different channels: through the app, phone calls and text messages. Since this is a useful tool, a minimum level of service (one advice Expected trend of # of requests by Category. Beginning vs End of Step 2. “ We noticed that you withdraw money from the ATM in Viale Monterosa daily on your way to work. There’s a closer ATM in Vicolo Corto with no commissions. Would you like to have directions to it? “ every two months) could not be considered as bothersome. More generally, each customer will be able to decide the level of proactiveness of the system. In this step, the new role of the call center employee becomes more prominent, so the personnel number is reduced over time, leading to an expected reduction of 70% at the end of Step 2, as most of the calls are handled by the A.I. The remaining operators will focus on the most specific and complex tasks. The A.I. introduction must be communicated correctly to people: a great innovation, if advertised in the wrong way can produce negative effects. The main focus of the communication strategy is to highlight not the A.I. itself, but the benefits that it will bring to the customer. Thanks to the beta testers it is possible to tell customers about real examples of how A.I. helped others save time and money. The target for the second step is to have the A.I. respond autonomously to 99% of requests, so the expected duration of this step is 18-24 months. In the third step the customer cannot request to speak with an operator. The A.I. may transfer the call to an operator if it is not able to solve the request. Now that A.I. has taken over the call center, the focus shifts towards the BUILD part of the new meaning. The A.I. backend is now a great asset for the company and can be used to build something bigger and more ambitious. Today in Mediolanum there are four types of accounts:  Conto Mediolanum.  Conto di Base.  Conto Freedom+ Professional.  Conto Young. They differ because of various parameters, just to name a few:  Yearly fee.  Deposit account interest rate.  Mortgage interest rate.  Trading features and fees.  Money transfer Fees.  Credit card fees.  Bonuses & Rewards.  Maximals.  Internet Payment Options. As seen in Chapter 4.2, each account type appeals to a specific audience and needs. The third step wants to radically innovate this concept by offering something completely new: Mediolanum now has the possibility to do what every company is dreaming of: one to one segmentation and one to one product customization. Today, the bank offers some kind of customization, based on the amount of money left in the account, but the goal here is to shape a bespoke account type for each customer from the ground up. This is the ultimate expression of the company’s slogan “Costruita intorno a Te”. This new service will rely on two main components:  Customer Service A.I. The A.I. backend that has been developed and trained for the customer service has expertise in the specific requests that the customer asks when interacting with the various channels. Here there will be information about the customers’ needs and behavior.  Raw data from Mediolanum’s Databases The raw data that Mediolanum already has about the customer and his activity (balance, expenses, movements, demographic information...). Here there will be information about the customers’ spending/earning habits and which services/features they use the most. Moreover, it’s possible to see whether they do for example online trading or have a mortgage. These two entities are combined into a unified Mediolanum Customer A.I. The customer will be able to create his Personalized Account Type after a sufficient amount of heterogeneous information has been gathered by the system. The system will also ask, right before creating the account type, some additional information explicitly to the customer. Here is a summary of the kind of data that Mediolanum Customer A.I. will have at disposal to create the custom account type. “Building an AI-generated bespoke type of account for each customer.” Obviously, the final goal of the bank is profit, so the final solution has to be profitable for Mediolanum. For this reason, the other part of the equation is represented by a “ Profit Level Engine ”, whose job is to calculate the profitability of a certain type of account with a certain set of parameters. The situation can now be reconducted to an Optimization problem, where algorithm will combine data from Mediolanum Customer A.I., maximizing customer satisfaction, services and constraints that come from the “Profit Level Engine” that will ensure the profitability from each of the account type. The outcome of the algorithm is a customized account type, with the features and fees that are most suitable for the customer, while keeping a defined profitability level for the company. Optimization Algorithm Max (Customer Satisfaction) Company Profit Constraints Profit Level Engine Customer Service AI (Needs & Behavior) Customer Activity (Raw Data) Mediolanum Customer AI Custom Account Type Transaction Logs Call Center Interactions App & Website Interactions ATM Operations Geographical Data and Location Online Shopping Data Preferred Payment Methods Most Used Services Additional info and preferences given by the customer when custom account type is created Mediolanum Customer AI Customers will be able to have their bespoke account type only after they had a certain number of heterogeneous interactions with the bank, to make sure the system has enough data and to generate a concrete offer that will bring real benefits to the customer. The average time that the customer has to wait to get his custom account type is one year. Moreover, the customer will have the opportunity to modify some variables and characteristics of his plan, with the “Profit Level Engine” making sure that the choice of the customer is still maintaining a sustainable profit level for the company. Customers can also choose to stick to their current plan if they desire to do so. This service represents a big selling point for Mediolanum, changing the Unique Value Proposition of the entire company. A dedicated marketing effort will be put in place by Mediolanum, using various channels, to promote the new service and attract more customers. The campaign will target the promising Millennial segment using online channels such as:  YouTube Ads.  Facebook Ads.  Instagram Ads.  Google AdWords. A marketing campaign using traditional media will be launched to target the Business/Freelancer segment. The channels used will be:  Billboards.  TV Ads.  Print Advertising on sectorial magazines on finance, business and entrepreneurship. Considering that in the actual conditions Mediolanum receives about 11.901.600 calls per year, so 32.607 ca. calls per day, the actual ratio between calls and employees is around 73:1. With the A.I. implementation, considering that the new technology will be able to answer to the 99% of the questions at full learning capabilities in Step 3, the contacts that will be transferred to a human operator will be equal to 326 per day. Since Mediolanum’s call center is considered very efficient, the new solution will keep the same ratio. According to this, the new number of employee for the call center is 5, with a reduction of 98,9%. All the other employees will be now moved into other offices of the bank where they can perform more complex tasks. In this section the journey of the customer through the various steps of the solution is detailed. Please note that it will describe the path only until the creation of the bespoke account type. Mockups of the Billboard Marketing campaign. Mediolanum App Shared AI Backend Following Mediolanum’s mission to allow the customer to bank wherever he wants and whenever he wants, the solution is touching almost all the channels. The A.I. has been integrated through the various channels keeping one single backend A.I. Engine. As exhaustively described in the solution the Call Center experience is completely revamped through the introduction of A.I. The app will now feature a convenient button to contact the A.I. Assistant and get things done quickly through voice. The integration has been structured through a voice-chatbot style conversation. When the customer enters his personal Area, a button will allow to activate M.I.A. As personal assistants become more and more used by people on Smartphones, Mediolanum will integrate its services in the most used ones (Google Assistant and Siri) through API (Application Programming Interface). Upon a permission is granted, the customer will be able to ask Google Assistant or Siri questions related to his bank account. The Request will be handled by the API and transferred to the Mediolanum App, which will then forward it to the Shared A.I. backend. The response will then be forwarded to the personal assistant. Through this channel the customer will only be able to perform non-sensitive operations. AI Backend Call Center Website App Siri + Google Assistant Integration The advantages of this solution are not achieved immediately, instead they will be reached over time step by step. This makes the solution not to be easily copied by competitors and so a competitive advantage overtime will be achieved. By creating the product directly over the customer needs, Mediolanum has an incredible advantage over other banks with “traditional” account types. Since this project aims to provide the best customer service without losing Mediolanum’s perspective, the solution is focused on an efficient completeness. KPIs are a fundamental step to understand if the implementation is addressing in the right way for both sides. According to this, the KPI’s balance-scorecard is divided in the following three main areas: general, internal and external. KPIs are not only aimed at monitoring the new solution performances but they should be a guide through all the process. The main scope is to push innovation even further and provide stimuli to employees. According to this idea, the new balance scorecard is composed by indicators, such as Avg. # of Innovative Solutions, that should encourage the people inside Mediolanum to ride the wave of innovation. In this sense, even A.I.’s Having a custom account type build for your needs is the ultimate expression of the company’s motto: “Built Around You”. The Call Center will always be available for the customer at anytime and fo r any kind of request. The Service offered by t he call center and the other channels through AI saves the customer time and money. The service is easy a nd i ntuitive to use and having a shared backend improves the familiarity with the tool across multiple channels. KPIs have to aims to: first, monitor the performances; second, understand where the service is more effective and carry on in that direction. Indicators distinguished by this symbol form the A.I. Scorecard: it monitors the evolution and development of the A.I. and the real advantages and benefits for customers and Mediolanum given by A.I. Indicators distinguished by this symbol monitor employees’ performance and satisfaction related to their role in the company. KPI Name Description Metrics Contacts Volume # of contacts for each channel per month # of phone calls + # of requests through the app and website + # of contacts with the family banker Avg. Talk Time Time spent on ph one calls and talking with the app per month (# of minutes spent on phone calls + # of minutes spent talking with the app) / customers Yearly Investment in A.I. Amount of Investment per year € invested per year # of Family Bankers # of Family Bankers # of family bankers Errors of Voice Recognition # of errors that A.I. makes in recognition customers through the app and phone calls # of errors in voice recognition Avg # of Accounts per Family Banker # of accounts assigned to one family banker # of accounts / # of family bankers KPI Name Description Metrics Value Added D ifference between the gained value for each useful information to profile the customer and the cost to obtain it. Value of the information obtained – cost per contact. % of Calls given to Human Operators % of calls that A.I. cannot solve and that are passed to a human operator. # of questions solved by human operators / total # of questions. Avg. Time for New Operation Time needed for the introduction of a new operation in the system. # of days for a new operation’s implementation / # of implemented operations. Forecasting Accuracy % of correctly forecasted requests. # of correctly forecasted requests / # of forecasted requests Motivation M otivation of employees regarding their own relevance for the company and for the customer. questionnaire about their importance in the company Avg. # of Innovative Solutions Avg # of new and innovative solutions and suggestions about financial and non- financial activities proposed by family bankers. # of ideas / # of family bankers KPI Name Description Metrics Avg. Time to Complete the Task Avg time needed to complete a task or request. # of seconds to complete a task / # of tasks % of First Call Resolution % of calls that A.I. completely addresses the caller’s needs without having to transfer, escalate or return the call. # of completed request during the first call / # of total requests Customer Satisfaction Customer satisfaction for the service. survey regarding family, personal bank account, ways to contact the bank, reclaims and critical issues Availability Availability of the system. # of hours with system ON per year / total # of hours per year # of Useful Advice # of useful advice to the customer considering the proactive approach. # of useful advice / total # of advice Empathy A.I. 's attitude in understanding customer’s feelings and behavior against uncanny valley. analysis of the tone of the voice during the call and the interaction with the app by the customer The solution presented is very difficult to evaluate in detailed monetary terms for a variety of reasons:  There is no off-the-shelves solution for an A.I. implementation like this one since pricing is decided upon a detailed technical analysis of the specifics.  Revenues are all indirect, meaning that there is no direct cash flow coming from the project, but it will bring value to the company indirectly in ways that are difficult to precisely estimate. Here is a projection of the costs, revenues and cumulative profit of the implementation. The values for the customer base growth are added to the growth that Mediolanum is currently experiencing (3.55% YOY). For these reasons, to each voice a range of values has been assigned, as described below. Cost/Revenue Range Range Values (€) Cost/Revenue Range Range Values (€) 1 1.000 – 5.000 5 100.000 -500.000 2 5.000 -10.000 6 500.000 -1.000.000 3 10.000 -50.000 7 1.000.000 -10.000.000 4 50.000 -100.000 8 10.000.000 -50.000.000 Cost Voices Range Cost Evaluation Info Co ding 7 The cost for developing Siri was 200 M $, since this A.I. has not all her characteristics and the numbers of customers is fewer, the estimated cost is around 20 M $. Nuance is the company that provide voice recognition and analysis Servers 6 Instal lation 7 IT Team 4 Energy 3 Qualitative estimation. Data Center 3 Online campaign 7 Marketing costs depend on different aspects and media. Estimated for a one-month campaign: on public transport around 5K $, on TV in the prime time 75 K $. Offline Campaign 7 Revenues Voices Range Revenues Evaluation Info Increased profits from existing customers 8 The revenues coming from the A.I.’s implementation have been estimated supposing an increment of customers and revenues deriving from the Profit Level Engine. (see Appendix for calculations) Customer Acquisition As said before, the economic analysis is provided just as a reference point. The economic effects are hard to measure and they will impact on the long term. The solution is surely a big investment for the company, but it should be considered not only for its economical gains, but also for the fact that it will give Mediolanum a big advantage over competitors in terms of long term innovation strategy. From the foundations to the final Build step, coherence and Manifesto values have been the guidelines through the process in order to reach a solid and consistent solution. The introduction of M.I.A. is sustainable and can grant value for both customers and the company. It is a huge source of innovation, useful to ride social changes and technological competitiveness, that nowadays are faster and faster. Through the A.I. implementation, Banca Mediolanum can make an incredible jump into the future: the perspectives of it are wide, with uncountable different applications, that in the future could be used even in other business of the Group (for instance insurance). From a financial point of view, the payback time is 5 years. It is a positive result, that is not considering the economic potential impact that the Artificial Intelligence will have in the following years in terms of brand equity, knowledge and culture. I see me as: The team sees me as: “ Having always team worked with people that I knew, finding myself in a completely new situation with new people is something thrilling. In my opinion Change and Failure are not threats, but opportunities to grow. This experience put me outside of the comfort zone, and that’s why I Ioved it. ” Enrico was the guide of the team, setting stages of the work that had to be done. He was the visionary mind, looking always beyond things. “ It was the first time in which I worked with unknown people . T his made me feel quite anxious but, when we started to know each others, it became a beautiful experience. This has allowed to obtain important results: to work in a harmonious atmosphere, to create positive relationships and to improve team-working and personal skills...” Nicole was detailed oriented and often gave critical observations to the project. It was fundamental to her to have the work finish in time. I see me as: The team sees me as: “ To summarize the project work in one word, I would say it was challenging. The challenge had two dimensions: the subject was very abstract and additionally team management was not easy. I think our team did a great job in overcoming all these difficulties to reach the common goal…“ Tayfun controlled that the work was done. Since he was the only foreign guy, he had an essential role in English communication and grammar check. I see me as: The team sees me as: “ From the beginning I was enthusiastic about the idea of a mixed group because I think it's an opportunity to meet new people and get to know different ways of thinking and seeing things. I am satisfied that, despite the difficulties encountered, we have been able to do a good team work. I have learnt a lot of things from my teammates…” Elena was meaningful for the positive motivation of the group, especially in the turbulence moments. She was precise and careful about the completeness of the project. I see me as: The team sees me as: “ The experience of taking a project from scratch and building an innovative solution on it, with a group of people that I’ve never met before, has been really constructive. I’ve nurtured relationships with teammates that have helped me in knowing myself better than before…“ Luca looked always to the big picture, he checked the coherence of the whole work. Since time was a constraint, it was important to spent it efficiently. I see me as: The team sees me as: “ It was a really challenging project, both because it took a huge amount of time and because of its difficulty. But at the end I'm satisfied of the result achieved. I really enjoyed being part of this amazing team… “ Alexsandro provided different resources, making possible a deeper analysis. His analytical capabilities have been significant. I see me as: The team sees me as: We modeled our project journey using the Tuckman Model for team development. Our adventure has been mapped through the four Phases, with a description of each one. We also mapped the performance of the group and the amount of issues/problems encountered overtime and we created a graph to map them, in order to have a clearer idea of what is going on during the various phases. The actual curves that describe the team’s productivity and issues do not strictly follow Tuckman’s model distribution, as they try to describe the actual situation that our team has experienced. At the beginning, no one knew the other members of the team: we were just different people that were expected to work together. This first phase was characterized by a lack of trust and, most notably, by having some problems in assigning and coordinating work, since we did not know each member’s strengths and weaknesses. In this phase the main problem was finding a collectively shared and right path to reach the goal. “ This project has represented a new challenge for me. It was the fi rst time I had been working with a group of new people using a different language. During meetings we had the opportunity to know each other and it became a beautiful experience…” Francesco gave useful advice, observing our personalities and behaviors. He tried to understand the best gear to make the team work. I see me as: The team sees me as: Forming Storming Norming Performing Productivity Issues/Problems As we began to have a clearer vision about the project, each member started to propose ideas, showing different ways of thinking and personalities between the members. This caused some conflicts among members, but we decided the path to take in a polite way, trying not to spiral the discussion into time consuming fights. Gradually we began to build our team, creating a sense of group identity, respecting and taking advantage of different perspectives and different takes on our project. We didn’t have explicit rules, but non-written ones have emerged over time such as:  Be present as much as possible during group meetings.  Important decisions are taken together, and if we are split we just vote. In this phase, we started to develop a strong commitment to the final goal and to really increase our efforts to reach it. This was due both to a stronger sense of belonging and to deadlines that were approaching rapidly. At this point, we knew each other and dividing and coordinating work, based on out each one’s strong skills, became easy. In the last part of the project we had less friction, as we knew that time was running short, so for the sake of the whole team whenever we had an argument we tried to solve it as quickly as possible, usually compromising. In the last period our workflow became very effective and with the same amount of effort we were able to do 4-6 times more than what we would have done at the beginning of the project. To conclude, roles, as rules, emerged over time, without the need of making them explicit. They become balanced and equilibrated: the free expression of each one was possible, letting them to give the best contribution to the team. Thank you for your attention, we really hope that you like our work as much as we liked to work together to create it. - Team 15  https://www.bancamediolanum.it  https://www.bancamediolanum.it/staticassets/documents/comunicazioni/2018/ BilConsolidato2017.pdf  https://www.multivu.com/players/English/8075951-teradata-state-of-artificial- intelligence-ai-for-enterprises/  https://www.accenture.com/it-it/insight-dynamic-consumers  https://www.forbes.com/sites/louiscolumbus/2017/10/16/80-of-enterprises-are- investing-in-ai-today/#63f846f34d8e  http://www.ninjamarketing.it/2017/04/19/ecco-i-vantaggi-di-una-strategia- omnichannel/  https://pages.arm.com/rs/312-SAX-488/images/arm-ai-survey-report.  http://www.pictetperte.it/business-e-innovazione/2017/06/12/il-mercato- della-robotica-cresce-la-lotta-tra-usa-e-cina/  https://www.idc.com/getdoc.jsp?containerId=prUS43662418;  https://www.corrierecomunicazioni.it/industria-4-0/industria-40-si-apre- la-partita-degli-investimenti-in-pole-l-automotive/  http://www.infodata.ilsole24ore.com/2018/01/16/boom-investimenti-nel- fintech-28-miliardi-solo-9-europa/  http://edition.cnn.com/2010/HEALTH/10/01/health.job.making.depressed/i ndex.html;  https://futureoflife.org/background/benefits-risks-of-artificial- intelligence/  http://theconversation.com/what-an-artificial-intelligence-researcher- fears-about-ai-78655  https://www.inc.com/jacob-morgan/should-we-really-fear-artificial- intelligence.html  https://www.mycustomer.com/service/channels/how-will-the- technology-mega-trends-shape-proactive-service  https://www.safehome.org/resources/privacy-and-technology/  https://www.nytimes.com/2008/01/03/technology/personaltech/03how.h tml?scp=7&sq=%22uncanny%20valley%22&st=cse  https://www.wired.co.uk/article/uncanny-valley-tested  https://www.wired.com/story/future-of-artificial-intelligence-2018/  Unheimlich. Dalle figure di cera alla Uncanny Valley, Pietro Conte, PSICOART n.2 2011- 2012;  R. Verganti, Design-Driven Innovation, Rizzoli-ETAS, 2009  R. Verganti, Overcrowded, Hoepli, 2018 Revenues # of Customers Revenues From Each Customer 60.330.000,00 € 1044000 57,79 € 1 2 3 1 2 3 1,03 1,05 1,07 59,52 € 60,68 € 61,83 € Hp: %of customers who will adopt the new account 80,00% AI 20.000.000,00 € 20,00% Marketing Campaign 10.000.000,00 € 30.000.000,00 € Natural trends of Mediolanum 1,0344 #Years # of New Customers Revenues Increase with respect the previous year Costs Hp: 1 year after the implementation of the AI 1,055 1 1079913,6 0 2.075.352,00 € 20.000.000,00 € 2 Years after the implementation of the AI 1,09 2 1117062,628 0 2.146.744,11 € 10.000.000,00 € 3 Years after the implementation of the AI 1,1 3 (implementation of AI) 1178501,072 9.406.920,47 € 5.184.824,36 € 30.000,00 € 4 1284566,169 16.870.950,24 € 7.464.029,77 € 30.000,00 € 5 1413022,786 25.897.522,88 € 9.026.572,64 € 30.000,00 € 6 1554325,064 34.520.275,17 € 8.622.752,29 € 30.000,00 € 7 1709757,571 44.005.302,69 € 9.485.027,52 € 30.000,00 € Sum of year 3,4,5,6,7 Increase of the number of customers After the implementation of the AI it will take 5 years to repay for the investment Years after the impelentation of AI Hp: Increase in Revenues Revenues from each Customers Costs