Big data and creating customer growth in the age of AI
Browse Blog Topics

Big data and creating customer growth in the age of AI

Companies in every market sector are tapping into remarkable amounts of data to gain insights and create engagement with their consumers. Using artificial intelligence technologies powered by that data, businesses can develop new analytics to try to understand customers’ preferences, behaviors, and actions. Business and technology trendsetters will discuss how they are using new data sources and A.I. to personalize messaging and engagement. How should companies navigate customer privacy and preferences? What are the best practices when building or scaling an analytics program focused on customer growth?

At the Milken Institute Global Conference earlier this year, Samsung NEXT President David Eun participated in a conversation on these topics with Diamond Resorts CEO Michael Flaskey, Wells Fargo CFO and Sr. EVP John R. Shrewsberry, Accenture Digital Group Chief Executive Michael Sutcliff, and moderator Sumant Mandal, who is Managing Director of March Capital Partners.

Sumant Mandal: Great, thank you. This is an interesting topic these days, especially if you guys have been watching what’s going on in the halls of Congress last few weeks, months.

And we have an interesting panel all the way from people who are at the cutting edge of technology serving customers all over the world, to people who are advising big corporations, other companies in how to navigate this world. As well as people who are the beneficiaries of using these technologies, and are using them to build their businesses.

So, maybe as a starting point, I’ll just ask David to start and introduce himself a little bit, why this is of relevance and why he’s here with us today. I’d love to know.

David Eun: Okay. All in two sentences. Hi, I’m David, I’m president of an organization called Samsung NEXT. We are like a startup within Samsung, and we are focused on startups. We are trying to find cutting edge breakthrough software and services to compliment our hardware. Because we believe in the future, it’s a thoughtful integration of hardware and software that will bring the solutions and experiences people need.

And the reason why we and I am focused on AI is, I believe in the same way that, to a certain extent, every company needs to be a technology company, because technology has absolutely changed the way it does business, the way it perceives itself. We believe that every company will be an AI company, because AI will fundamentally change the way it goes about doing it’s business.

Sumant Mandal: What David didn’t tell you is that he and I have been in this technology business more than two decades, at least. And we both invested in the same entrepreneur, not the same company. And we both lost all our money investing in that guy. And we just found that out this morning.

David Eun: But amazing learnings, amazing learnings.

Sumant Mandal: Yeah.

Michael Flaskey: Good morning. I’m Mike Flaskey. I’m the chief executive officer of Diamond Resorts. Diamond Resorts is the world’s largest independently branded vacation ownership company. Big data and AI is critical in everything we do. Really three primary phases that we focus on. The customer acquisition, the customer journey. And what I’ll talk to you more about today throughout the panel is, how we through innovation have taken experiential vacations, and we’ve redefined our industry through the use of big data and AI.

Sumant Mandal: Great. John.

John Shrewsberry: I’m John Shrewsberry, I’m the CFO of Wells Fargo. I also manage the technology team at Wells Fargo. We are a seven to eight billion dollar a year technology spender, and data is at the center of all of that. Whether it’s from the customer experience and customer acquisition side of the business, or from risk management and fraud management, and how we maintain the transactional records of 70 million customers doing the things that they do through their banks.

So, this is a hot topic. We probably have a hundred or so AI proofs of concept, or in production models today. And every part of our business over the course of the next year or two years, will begin to change as a result of the smarter use of data.

Sumant Mandal: Now, I’m not sure if Mike, if Wells Fargo is one of your customers or not, but maybe you can tell us-

Michael Sutcliff: Absolutely.

Sumant Mandal: … a little bit more about what you do.

Michael Sutcliff: Yeah. So, I’m Mike Sutcliff from Accenture Digital. We are convinced that artificial intelligence is kind of the alpha trend in technology. So, it’s impacting every part of our business. We have Accenture Interactive as part of Accenture Digital. It’s just been listed again this week as the largest and fastest growing digital marketing network in the world.

So, we use a lot of big data about customers to help our clients create growth in their business. But we also have a group called Applied Intelligence, which is thinking about how we apply this artificial intelligence technology to every industry that we serve. From mixing chemicals in specialty chemical plants, to a wide range of different solutions across industry.

So, for us this is a big business. We’ve got about 3,000 data scientists working on it full time, and that’s growing very rapidly.

Sumant Mandal: So, given the present environment around this panel’s conversation around the consumer or customer data, what has been the approach at Accenture as you work with many of your clients, in thinking about how they balance usage versus privacy? Maybe we’ll start with that, and then go deeper into understanding how and what people are actually doing in some use cases of the companies here, and how they use this data for the kinds of applications that they are serving.

Michael Sutcliff: Sure. Well, I mean, first of all, our arching philosophy about artificial intelligence is that we should be developing what we call responsible artificial intelligence. That we should understand the data, where it comes from, what permissions that there are around the data in terms of how it’s intended to be used. We have to understand the models, and whether there’s intentional or unintentional bias built into the models. And then, we have to understand how it’s actually being applied.

We actually just released a book, Paul Daugherty, from our chief technology officer just released a book called Humans + Machine. So, we’re looking at how humans-

Sumant Mandal: Good book?

Michael Sutcliff: Yeah, humans and machines come together to do things that neither one of us can do alone. So, we’re really focused on understanding responsible AI at a broad level. And then, working with our clients to use it for lots of different things. Sometimes it’s at a population level where we don’t need to know an individual’s name. But what you’ll hear I think at Wells Fargo, is you’re using it for your customers, and it’s a very individualized response that you’re giving.

So, we see broad and narrow uses. And it’s really interesting because we can do something every three or four months that we weren’t able to do three or four months ago, because the technology and the tools are changing so quickly.

Sumant Mandal: And so John, where you’re sitting, which is in the, I would say, financial management, risk management part of the business. How do you think of these technologies, and what level of engagement involvement do you have in trying to guide the company in what they can and can not do?

John Shrewsberry: Sure. Well, it comes in a few categories. One is, the company should be thinking about data as an asset. The company should be thinking about pockets of data that are generated all day, every day, everywhere, and bringing it together for the benefit of the customer. So, to help customers make better decisions to avoid costs, to increase returns, to save a little bit more and prepare for retirement, to avoid fraud or identity theft.

Incidentally, one of the troubling parts from a bankers perspective of what’s going on with the focus on data privacy is, our customers themselves often do less than what they should to protect their own data privacy. Just as an example, I’m sure there are a bunch of people in this room that would have given their username and password to their bank accounts to apps that they’ve downloaded on their phone, to aggregate information, and present it to them in a colorful way. And meanwhile, they’ve given their username and their password to their bank account, to an app on their phone. That’s something people need to get comfortable with-

Sumant Mandal: So, very early in this sort of trajectory of adoption of these new technologies, and there’s lots of room for error on both sides. David, how does Samsung think of data? How do you think of data?

David Eun: Well, the first thing that we always remind ourselves is it’s the first inning of a nine inning game. So, it’s still very early days. And while we’re trying to figure out what the foundation and the, literally, the tools for developing AI are, that tsunami of data continues.

So, right now there’s something like eight and a half billion connected devices. Not just smart phones and TVs, but a typical car has something like 12 to 30 sensors in it. And as we think about new businesses at Samsung, like autonomous driving, or AR and VR, or connected homes and security, the proliferation of data will just increase.

We think a lot about how it is that infrastructure and smarts at the center in the Cloud will evolve. And we believe that that evolution will include having much more interactivity and connection with things closer to the edge. So, Node’s at the age, and in fact, everyone I presume has a smart phone in their pockets. That is now a camera, sensor enabled super computer. So, while it is documenting and creating lots of data, we believe it has the ability of course, to store and also process information.

So you’ll see in the following innings, the ability for individuals to use their smart phones, node’s closer to them, perhaps in your home or in your car, to process and determine what kind of data needs to be curated and go to a Cloud, and what kind of data needs to be processed and decisions made there.

If you’re in a autonomous vehicle and you’re driving 65 miles an hour, you may or may not want to ping a Cloud and have it come back to you because of latency or connectivity issues. And you may need for that car to swerve to avoid an obstacle. So, there will be more and more instances in the future as we look to improve our lives and as technology continues to evolve, that you’ll have to figure out what type of hardware and software needs to be more powerful closer to you as a user, and how much of it resides in the Cloud.

Sumant Mandal: And I think one real take away from here, and I’m sure people in the room already understand it, that it’s not just your username and password that you have to protect. Every different action you take in the day, every thing you engage with, there’s some “data” being generated, and somewhere being stored. And some question around, who does that belong to?

So, Mike, you run a company. You have lots of customers that use your product. And the way you describe your product is experience. How do you think of these new technologies, and how do you as a CEO think of this as being part of a business?

Michael Flaskey: Sure. Well, there’s three primary platforms for us. On the customer acquisition side, in 2017 we had 300,000 families that sat face to face in a direct sales presentation learning about a better way to vacation. And in order to get those 300,000 guests to sit in front of our sales people, we have a whole team of people that focus on using AI to make sure they get the targeted right people in front of the customer.

So, on the customer acquisition side, we’re touching millions and millions of people that are feeding us this data, that allow us to get a customer in front of our team members that are much more likely to buy. Which overall, saves marketing costs.

Sumant Mandal: And have you seen a dramatic change in how customer acquisition works once you’ve used these technologies?

Michael Flaskey: We have. And the most interesting part of it was, we had a revelation about three years ago. If you go out and you look at the data, what the data tells you about our industry is, we have over a 90% … People love the product. People, when they use our product and they vacation at our resorts, they come back blown away. They don’t like the sales process. People don’t like the direct sales process.

So, we used this data, and we created an experiential platform in 2013, that has really redefined our industry. And today, about 45%, or in 2017, 45% of our sales came from an experiential platform that AI provided us the technology and the data to be able to go out and target for the customer.

Sumant Mandal: And when you’re looking for these customers, potential customers, are you outsourcing that to some third party to go find you-

Michael Flaskey: We do it all internally.

Sumant Mandal: You build your own technology, you use your data internally. You don’t farm it out, you don’t borrow data or buy data from other people?

Michael Flaskey: That is correct. We have a homegrown platform called Diamond Clarity. And the technology piece that really drives the customer acquisition piece is what we call our propensity.

Sumant Mandal: So, what one sort of takeaway from here is, you could be in the vacation business, you’re really now a technology company.

Michael Flaskey: Without question.

Sumant Mandal: And Mike, coming back to you, what do you see in this transition in industry? And I’d love to hear from you as well John. And this sort of, how do you see technology being appreciated, these specific technologies now that they touch, I think every part of every business, and what’s the values here?

Michael Sutcliff: Well, the interesting part is, if we think about Accenture Interactive, which is our marketing agency. We actually don’t describe it as a marketing agency, that’s the term the industry uses. We actually talk about an experience agency. Because the world of marketing has transitioned from broadcasting messages and creating basic awareness, to creating a brand promise around an experience that you’re going to have. And then the question is, can you deliver that experience?

And so, our perspective is that we’re going to be using technology over the entire customer life cycle. And the better you can get at providing personalized experiences, the more value you’re creating for your customers. But sometimes the technology is invisible, and I’ll give you an example.

We just did a piece of work in Japan for Toyota. And Toyota said, “Look, we want more taxi drivers to choose Toyota as their car.” So, we want to create growth for the Toyota brand, specifically in the segment of taxi drivers. And we said, “Okay.”

We went to KDDI, which is one of the large telecom companies, and we said, “Can you give us anonymous phone records, we don’t need to know anybody’s name. Can you just tell us from the call records where people have dialed cabs over time for the past year?” And they said, “Of course,” ’cause we know the numbers of the cab companies right?

So, we can just lift out the records out of all the calls that were made in Japan, let’s just look at the ones that were called to taxis. When were they made? Where was the person standing with their phone? How was it related to events going on in Tokyo, et cetera?

Then we did population level analytics, and we forecasted for the taxi drivers driving Toyotas. In the next five minutes, where is the person closest to you going to be calling for a cab? So, how can you position yourself closest to the next person that’s going to need a cab?

Now, the person who dials doesn’t know that that’s happened in the background. But there’s actually artificial intelligence and machine learning being used to constantly predict where the next event is going to occur. In this case, somebody’s going to call a cab.

In our pilot, which we’ve just finished, an eight hour shift has seen a 30% uplift in the taxi driver revenue because they’re not in the wrong spot. As soon as they finish this, we’re telling them where to position for the next one. And they’re actually driving fewer miles, so they’re saving money.

So, we can generate revenue uplift for a taxi driver. We can save them money, ’cause they’re not wasting miles driving somewhere where there’s nobody looking for a taxi. And that creates revenue growth for Toyota.

Now in that context, there’s no individual who knows what’s going on in the background. It’s just that everybody gets a better answer.

Sumant Mandal: John, how old is Wells Fargo as a company?

John Shrewsberry: 166 years.

Sumant Mandal: How does a 166 year old company think of technology?

John Shrewsberry: Well, as essential.

Sumant Mandal: And one more question for you before you do that.

John Shrewsberry: Sure.

Sumant Mandal: What’s the average age of the employee in the company?

John Shrewsberry: It’s bimodal. I’m going to say it’s 30ish, but there’s a big concentration-

Sumant Mandal: Relatively young.

John Shrewsberry: … of young, and a bigger concentration of older.

A lot about banking has changed in the direction, or as a result of technological breakthroughs over the last 10 or 15 years. From the ATM to mobile banking and other things. This harnessing of data in a way that provides more customer insight, more customer convenience is new.

So that it’s that much easier to become a customer. You don’t have to go into a physical location. All that can be done through a combination of the attachment to a phone device, and a few things that can be told about you. That’s a real convenience right there.

Sumant Mandal: The money becoming electronic.

John Shrewsberry: Money is increasingly becoming electronic. There’s still a lot of cash in the system, but small dollar payments are easy to do in either a bank owned way, or in other ways. We’re really trying to make it convenient for customers to … For the bank to know who they are.

So, what irritates some customers is if they want to apply for a loan for example, and they get a list of things to provide. They may say, “Well gosh, you know that about me. You’re my bank.” And that’s been done away with.

We have 70 million customers. If you were to apply for a mortgage at Wells Fargo now, everything we know about you from wherever that connection point is, we populate it.

Sumant Mandal: So, data silos are disappearing.

John Shrewsberry: Silos are breaking down, which is very important. It’s an efficiency driver for us, and it’s a huge customer experience uplift.

So, we’re using it to improve the experience for the people who are already our customers. And then of course, to make it easier to attract people as well.

Sumant Mandal: And is there a specific ask you make of your customers, to allow you to use their data? Is it your data, is it their data? How does that balance get created within an organization?

John Shrewsberry: Sure. Well, when you’re applying for a banking product, you’re definitely clicking through terms of use that generally speaking, say that we will keep their data private. We don’t monetize-

Sumant Mandal: And is that regulated in your industry?

John Shrewsberry: At the federal level, there are norms that are adhered to. I think you could enter into a different agreement with a customer if they chose to. It’s our approach that it’s a slippery slope to go down if what we know about customers begins to be merchandised to third parties, which is topical these days.

There’s plenty for us to do for our customers, with our customers, and ways for them to benefit with the data that we have.

Sumant Mandal: And is it generally accepted belief, and I don’t know, and this is purely [inaudible 00:18:20]. Is regulation around customer usage or data usage of customers, is that a good thing?

John Shrewsberry: I certainly think that it is. And of course, it’s stricter in different parts of the world versus where it is here. I think it’ll become an even bigger topic as people want to know that there’s control over their financial lives.

I think this pact that people have entered into with social media and others, where it was clear that they were allowing themselves to be marketed to in exchange for the convenience or the experience. I’m not sure where that goes. That’s at the heart of what we’ve seen recently.

Sumant Mandal: We’ll talk about that. I think that’s not clear to even the social media companies, to be honest. It’s that early in the generation of, evolution of this.

David, how many customers does Samsung think of as their customers? How many people in the world?

David Eun: Well, we have so many different lines of business. So, just at electronics we have a components business that supplies Samsung, but as well as other companies.

Sumant Mandal: Just focus on consumers. I’m just … Yeah.

David Eun: Well, a couple of facts. We sell something close to two TVs a second. We sell our smart phones in over 200 different countries. And when you think about the distributed effect of digital appliances, and now we have software companies like SmartThings, which is a platform to connect not just our things, but third party products.

We think increasingly about basically hundreds of millions of people potentially. And we play this interesting role also, where we’re a partner to different software companies, different apps companies, different social media companies. As well as carrier networks and big box retailers all over the world, who are all trying to form their own customer relationships.

Sumant Mandal: Which is what sort of leads my question, which is, you’re at sort of the physical level touching a customer. You’re at the operating system level, then there’s an application, “Facebook” like applications sitting on top of your operating system. And then there’s something else that’s using Facebook’s data to sort of market it to the customer.

How does this chain get resolved in a way that the customer knows that they have control over what they have, and it’s theirs? And in some way, how does a company like yours think of their responsibility in that value chain?

David Eun: I think it differs by company, it differs by customer, and it’ll differ by time. It’s probably tempting but inappropriate to supply a one answer fits all, just in a situation like this. But we believe in a clear communications or transparency to the customer is essential. And giving customers choice about what they want to do.

And this is why, when I spoke of edge devices earlier, giving these edge devices the ability to communicate and give users choice, along with these partner companies to say, “Hey, this kind of data has been gathered about you. This data is being shared in these ways, and here are some choices.”

So, there is a dial as it were, a virtual one, that customers will want to toggle to say, “Hey, I may want more personalization, and I’m more comfortable with more of my data being shared, whether inhonestly or not.” Or, “There are certain parts of my life that I would not want to share, and want to have closed.” And increasingly, users will have the ability to do that. And we think that there’s an opportunity and a role for us to help.

Sumant Mandal: Yeah. On the operating system, or-

David Eun: On all up and down the stack. So, if you think about it, let’s say you have a connected home and you have security. The security camera on your front door that’s connected to, let’s say a TV in the living room. You should be able to, through machine learning or AI, identify who those people are. And they may or may not be granted automatic access into your home.

That can be done, again, at a local level or it can be done on more of a systemic basis through a service, or a third party service. These are all choice that people will have, and all up and down that stack, we either ourselves or through partners will help to find what those choices are.

Sumant Mandal: So, data privacy, sovereignty of data, all the issues that we’re talking about are really issues on, I would say, the global level. Not necessarily on just a company or … I was in India last week and I was … We have a company that does financial services there, and they have this new system where 1.2 billion people are being put into a database. Iris scan, fingerprint, et cetera, et cetera.

I asked, “Aren’t you afraid of your data?” Because I can just Google anyone and get their bank account and what they have, and it’s on the web. And they said, “No. We’re just excited someone knows we exist.”

Sumant Mandal: So, this is eventual-

David Eun: One thing that people don’t talk a lot about in the debates and in the headlines is, I think there is actually a pretty stark contrast, even within our country, of people’s … Where they fall on the spectrum, if there is one. If there is a tension between personalization and privacy by frankly, age and familiarity with technology.

And I think as more and more that comes out, I think again, technology companies will have an opportunity of, not an obligation, to give users again, more choice in how they want to handle that.

Sumant Mandal: So, maybe Mike, that’s an interesting point to talk about in your business. What’s the demographic of the customer today, and how do you think of five year, 10 years, and what changes using these kind of platforms?

Michael Flaskey: Today, the demographic of our customer is 55 to 75 years old. You’re talking about middle America, combined household income of 90 to 150,000 dollars. Typically, empty nesters. You’re talking about two county workers, a police officer, a firefighter.

But its’ shifting. The world of the Airbnbs, the millennials, they don’t want the longer term commitment. What it’s forcing us to do through our AI platform is, it’s forcing us to look around three corners, and start to prepare for that next generation that’s coming along.

And what we know is, is we know that we’re going to have to define our product into more a term product. The millennial group, they do not want a perpetual product that we’re working now. In fact, we just tested and we’re getting ready to roll out system wide, a product that is a 10 year product. Where at the end of 10 years, the consumer can walk away from it.

And so, we talk a lot about machine learning, and we talk a lot about big data, but we still believe leadership plays a key role. And we believe it’s more about the velocity of how you collect data, and how quickly you can get it in the hands of a competent leader.

We think we’re decades away from machines running companies. And so, I’ll give you a perfect example that will probably be relevant to many of the folks in the audience. I was in the baseball industry before this, and in Major League Baseball when a manager goes in to interview, a potential new manager goes in to interview for a club, one of the main questions that comes out of the front office now is, “Would you have an issue if the front office used data to make the lineup, and they sent the lineup card down from upstairs?”

And the reality of it is, is that that’s a defining question in today’s age in Major League Baseball. And what you find is, is you find data is relevant at every point in every decision that is made. But ultimately, there still needs to be a field manager with the finger on the pulse of the people. Because what the data doesn’t tell you, is it doesn’t tell you how that person feels today. It doesn’t tell you what’s happened necessarily over the last 24 hours. It doesn’t tell you what went on in their home life.

And so, we believe strongly that we’re at an inflection point where it’s a mixture. It has not gone so far the other way.

Sumant Mandal: Maybe David, do you agree with that?

David Eun: Absolutely. Absolutely. I think-

Sumant Mandal: I would hope not, but okay.

David Eun: I think that the whole sci-fi machines will take over the world makes for interesting headlines. But it’s probably a little overblown. And we have to be thoughtful about it, but again, because it’s so early we can absolutely figure out ways to integrate the involvement of people in the decision making.

And if you look at so-called AI today, by the way, there’s so many different types of AI. But if you look at AI today, it really hasn’t progressed to the point where it learns like a human might. Where it has judgment like a human. And you know, we’re not even talking about ethics and things like that.

So, this idea of trying to capture information and process it like a human, is maybe a holy grail, but we’re not close to it. And so, for a long time yet, we will have to be involved.

Sumant Mandal: And Mike, how does Accenture think of that?

Michael Sutcliff: Well, the first thing is, we agree broadly that the different between generalized artificial intelligence and narrow band artificial intelligence is a significant difference. They’re two completely different topics.

Sumant Mandal: There’s no singularity coming?

Michael Sutcliff: No. No singularity in the near future. But we also can prove that a lot of things that managers believe that they’re good at, they’re not actually very good at. And that the systems can be more accurate in understanding, and in fact, applying what we would describe as judgment.

Now, we’ve seen that even in medical diagnosis, where we’ve proven that … And you’ve seen the articles about artificial intelligence being able to read radiology reports more accurately than radiologists who are making a half million dollars a year, deeply trained, specialized experts, in something that they’ve spent 20 years learning how to do. And yet, the technology is already leapfrogged and become better at reading those exams than they are.

That doesn’t mean that we don’t need radiologists. But it means that radiologists need to adjust what they think they’re good at. And the combination of the radiologist and the machines can be even better.

So, that’s why we address this topic of humans plus machines. We want to keep going through the learning curve to understand where and when the machines can assist humans in being good at what humans are actually good at. But we also want to recognize what machines are good at, and how we can let humans work with the machines, when the machines are the ones providing the incremental value.

So we see both sides of the coin, and we think we are in the first inning of a long learning streak.

Sumant Mandal: And so, John, for your organization, where do you see something like that actually effective, where humans are willing to adopt help from a machine?

John Shrewsberry: Yeah. So, there are a variety of use cases where if we’re trying to spot anomalies in big bodies of data. So, think about-

Sumant Mandal: Fraud.

John Shrewsberry: … money laundering, fraud, et cetera, where you’re looking at millions of transactions and looking for things that fit a certain pattern, that modern data analysis tools that learn, and adjust, and can be trained are invaluable to the people who want that raw material to work with.

There are other people however, where some of these tools is moving their career cheese, because it was their job to do that. And now it’s either more effective, or more efficient, or both to do all of the groundwork automatically, and then to deliver ten times as much actionable information to one person to make them more productive. That’s fantastic for some people who receive that, and that’s threatening for others.

Sumant Mandal: Is it cultural? Is it demographics?

John Shrewsberry: It’s human, I think.

Sumant Mandal: It’s human. What I read was, where machines are replacing humans is three types of jobs. It’s dull, dirty or dangerous. Dull, which just means it’s repetitive, and the machine can … Dirty, that human doesn’t want to do it, clean sewage or something like that.

John Shrewsberry: I would add to it also, jobs where people’s long held beliefs about their own professional judgment turn out not to be as effective as Excel, frankly. Let alone, the tools that we have-

Sumant Mandal: Now, we just heard from Mike here, the concept of ownership and the kind of change that has been brought on by the millennials, sort of, I think, influenced by technology, influenced by the availability of assets to rent, immediacy of assets. How does that play into your business?

John Shrewsberry: It’s certainly possible that either … Well, it’s sort of the employee side of the equation, and they’re probably a little bit more fluid in terms of their careers. Which is similar, sort of a gig mentality.

And then, there’s the customer approach which is, maybe they don’t have the same loyalty to one entity, but they want the ability to get some of this in one place and some of that in another. And enabling that where it’s appropriate, making it easy for them to understand what the possibilities are.

Sumant Mandal: Has the demographic of a Wells Fargo customer changed?

John Shrewsberry: Not so much. It’s beginning to. The pace of mobile interaction with customers is up steeply, but the fall off in physical interactions with customers has not dropped off at the same pace. What’s happening is, they’re interacting more frequently electronically.

Sumant Mandal: Which is good.

John Shrewsberry: Very good. A lot of engagement. But maybe because we still have 10,000 physical locations, they use most of them.

Sumant Mandal: That’s probably not true globally. I don’t know about Wells, but in general what I hear when you travel the world is, the physical retail location is definitely under threat, whether it’s gone away completely or not is under threat.

David, you have a portfolio of companies that you’ve invested in. And within Samsung you wear many hats. Talk a little bit about what your investing in, and why? And I know there’s some overlap with our conversation here, but what is it that’s interesting for Samsung as a next evolution of these technologies?

David Eun: Well, first it’s contact style. For the last three, four, five years we’ve spent over 12 billion dollars a year on R and D. So, as a big company we’re very much invested in identifying new breakthrough technologies.

But with regard to software and services, the reason why my group exists is because historically, innovation has tended to come from startups. So, the reason why Samsung NEXT exists is to compliment and then cover the entire spectrum of potential innovation, and this is also true for AI.

There are really interesting investments and things that big companies are doing, including my own. But so much innovation, and so many different pockets of development are occurring right now among startups, that we spend a large amount of our attention and focus-

Sumant Mandal: What’s pushing the envelope on big data and AI in your mind today?

David Eun: Well, any variety of things. But for example, one of the big areas that we’re focused on is, those who are building the tools are creating and developing tools for AI. So, we’re looking for people who are building the picks and the shovels for other people to then go and find the AI gold. And we think that’s a critical part of where we are as an industry.

So, many of our companies, we have a company called Vicarious that is creating software for robots to get more efficient. And so, it’s not applications or services around that, but just helping them become more efficient.

In some ways, when we look at AI now where we feel like it’s before the world wide web became what it has become. Before it was really just a very somewhat arcane thing for top research scientists, and government, and universities to access. But then came standardized languages and protocols. Then came TCPIP, and java script, and page rank, and these sorts of things that made it accessible to developers, who then made it accessible to users.

And we believe we’re not quite there yet in the AI world, but we are looking for those types of opportunities.

Sumant Mandal: I think one thing that we ignore when we talk about all of these technologies is the availability of data, because without data it doesn’t matter what tech you have. You can’t compete. So, if someone had to compete with you today, Michael, what’s a threat to your business in terms of, you have historic data, you have a brand, you have loyal customer base, you have … What do you think of when you think of these technologies?

Michael Flaskey: You mean from a competitive front?

Sumant Mandal: Yeah.

Michael Flaskey: Well, I think that the next evolution of customers is frankly, the biggest competitive threat we have. We don’t really have a competitor in the sense of someone across the street competing for our customer. What we have a need for is to develop a product that is going to make our industry sustainable long term as things evolve. And that’s where the AI has really helped us, as I mentioned earlier, at starting to develop products that are more suitable for the millennials.

Now, today the millennials, they don’t make up a large percentage of our customer, because they don’t have a lot of money. But we know it’s coming. And so, that’s what I would see as the threat.

And then the other piece of it is, is that it’s the customer journey. At the end of the day, you look at the customer acquisition piece that we’ve talked about, and you ultimately are striving to get that loyalty out of them. So, in order to do that, you have to be able to create aspirational product features that they want to continue to aspire to have, and they will remain loyal to your brand. So, the data feeds us that. And ultimately, what brings it full circle is the retention piece.

Sumant Mandal: I’m going to ask for an instant consulting here. What should he be afraid of?

Michael Sutcliff: Well, it’s really an experiential challenge, because the perception that people have in dealing with any part of their life transfers, in terms of their expectations and what it’s like to interact with any company. And so, they expect the company to know their preferences, their behaviors, and give them personalized responses.

And so, from our perspective, it’s really understanding the perceptions and the expectations of those customers and being able to serve it, even if it’s not something that’s standard in your industry.

So, most industries have fairly predictable disruption. In fact, we just produced a disruptability index. We looked across 98 industry segments, and we tried to understand how much disruption has occurred and how much is coming. And when you start to look at customer expectations and friction points in the industry, you can very reliably say if there are still a lot of friction points, there’s a lot of disruption coming.

Sumant Mandal: Okay. We’re going to shift a little bit, and I think we are going to have probably 10 minutes towards the end, five, 10 minutes for questions from the audience. But lets talk a little bit about what we saw with the whole Facebook customer data leakage, and a little bit about regulation that’s following it, especially from Europe.

I don’t know if David, do you have any experience or thoughts around how you see that evolving, and what does that regulation mean for your business, as well as businesses around you?

David Eun: I don’t have really any specific thoughts on regulation of this space.

Sumant Mandal: Let’s just talk about the regulation, is broadly saying that if someone uses a customers data, they have to get permission from them. And in this case, it’s not clear whether the permission is someone who’s, the publisher has to ask for it, or the advertiser has to ask for it, or the intermediary has to ask for it, or the person paying for it. ‘Cause there’s so many people. It’s still very early. So, I’m just curious as you think how that evolves.

David Eun: Well, I think it goes back to what I was talking about earlier, and we’re already beginning to see this, but increasing the transparency and communications with all those parties, with the user. And educating the user about what kind of information is being gathered and stored, and how it’s being used.

And again, I will say there is a very diverse body of opinion and feeling about what happens to my data as a user, depending on the user. And so, if you are a company that deals with this very diverse and large group of users, you’ll have to be very flexible and nimble in the way that you interact with them. Because at the end of the day, there is still tremendous value that comes from sharing my information as a user.

So, it’s not so much the gathering of that data, but what you do with it. And I think sometimes those thing become blurred. So, as a user, I don’t mind if you take my information, as long as you respect it and protect it. And I think for the most part, the companies that have been in the spotlight have had good intent, and have tried to do that, but the communications haven’t, by their own admission, been as clear as they would have liked. And I think you’ll see a lot more of that.

Sumant Mandal: It’s really just, in your mind, evolution then.

David Eun: As we said, it’s our first inning.

Sumant Mandal: Does Wells Fargo use social media for customer acquisition?

John Shrewsberry: We do. We have run chat box in both Facebook and now in Apple Business Pay. That’s a way to interact with customers in those venues. We do digitally market directly to consumers in social media platforms. We don’t provide information back to those platforms about people’s banking activity.

Sumant Mandal: Okay. And has that thinking changed any ways in the last 45 days?

John Shrewsberry: A little bit actually. Some of the, this may be banking specific, but some of those platforms have unusual sort of … The absence of contractual relationships. So, in terms of who is liable, who’s at fault, and in the case of bigger legacy companies, who can afford to pay. Those have becomes really big issues.

Sumant Mandal: Yeah, you have a lot to lose.

John Shrewsberry: We have a lot to lose. Most companies, you just go bankrupt. Even though they don’t want to, but it just works out that way. So, we’re thinking more deeply in all of these relationships about the risk of loss, about liability, and clearly, about the transparency of the business practices, the compact if it’s not contractual, that these services have with their users.

Sumant Mandal: And Mike, in your sort of world of companies you work with, how is this being perceived? What is changing? How do you advise people in how you think about the next 12 months, 24 months?

Michael Sutcliff: Well, we do work across pretty much all the geographies in the world. And as you said, there’s a wide body of thought on what’s appropriate and not, just by cultural, by age, by industry. So, we really start with transparency, permission, very clear permission.

But we also remember that the role of regulators is to protect the minority. So, if you have a vocal minority that says, “We don’t want this to happen,” then the regulator has a responsibility to protect that minority from that thing happening. And in any area of life.

I think if we look at what’s happened in Europe with their global data protection regulations, they’ll probably find that the vocal minority created a constraint set that the majority is not going to enjoy. And so, there’s going to be some back and forth when they start to realize the unintended effects of some of the data protection policies that they’ve put in place.

So, it’s a conversation. It’s a societal conversation. Each society has different perspectives. I was just in Stockholm last week, and I was in Brazil the week before. You know, quite different thoughts about what’s appropriate, what’s not.

Sumant Mandal: Great. We have, and the panel talked a little bit about how these technologies are being used, how there are some issues that we need to still resolve on the evolution of the technologies. Let’s talk about the future, and let’s talk about in terms of, especially as David, you think of your portfolio, or the portfolio of products within the Samsung network. What’s the future around a customer’s engagement with the network?

David Eun: Well, I think the customer will rely on their end device a lot more, because it will become increasingly powerful and knowledgeable about you as a customer. I think people will also invest in and understand that there’s a lot more AI built around them and closer to them, and not just in third party companies that have invested in building large proprietary stacks.

So, you might have a home that is smart and records, and stores, and process lots of information and intelligence. You car may have a lot of information and data that’s particular to you, that will have that.

I’m wearing a Wearable today. It’s constantly monitoring lots of different health functions. It’s constantly gathering information that is being streamed to some place right now. In the future, I may direct it to different places.

And the type of data that we generate will change. Not to digress but, back in the old days, data was relatively one dimensional, right? Rows and columns. And you had relatively flat software to analyze it. Today we have more textured data. You have video and text itself, and it’s not just rows and columns. In the future, there’s going to be 3D video, computer simulations, streaming health information, all sorts of AR and VR information all transpiring-

Sumant Mandal: With 5G network.

David Eun: … with 5G networks and everything. So, you will need more computing power all around you. And it will be embedded, and it will be invisible a lot of times. But it will be embedded everywhere you go. And so, those companies and businesses that embed this kind of power closer to users, will give people a better experience.

So, I’m going to talk to Mike later about making sure all his vacation homes are smart homes, so that the users will have better experiences.

Sumant Mandal: The sales pitch coming.

Michael Flaskey: I understand.

David Eun: But those are the types of things that I think we should expect.

Sumant Mandal: Yeah. Maybe from your point of view, how do you think of that evolving?

Michael Sutcliff: I don’t know that we know. As I think you said, we’re so early in the process, and we have so much to learn from so many different parts of the market. I think we’re open minded and we’re kind of early in the learning curve. So, I honestly don’t think we know the answer.

Sumant Mandal: I’m going to ask you a slightly different question then, because I do want some answer. There is a school of thought that says the big internet companies today, the five or so that matter in the world, have almost moats that are impossible to breach. Because they have so much data around their consumers, and there’s hundreds of millions of consumers. There are billions. Do you see that change in the next five to 10 years.

Michael Sutcliff: Nope. I think they have an advantage, and they’ll keep the advantage. Now, having said that, let’s just take what’s going on in medical research today, and talk about regulation versus impact.

So, one of the big debates now is, is China going to leapfrog both Europe and the US in the pace of innovation in medicine? And the answer is simply, yes. Because they’re taking all of their medical records, anonymizing them, and making them available for research. We’re not.

Sumant Mandal: Regulation is an issue.

Michael Sutcliff: Yeah. So, in this case they have a distinct structural advantage that will fundamentally change their healthcare market at a rate that’s way faster than we will. And by the way, we’ll be way faster than Europe.

Sumant Mandal: Yeah. And to some degree they need to, ’cause they don’t have the legacy stuff to protect as much as the US does. As well, the systems and services that exist today.

Michael Sutcliff: Well, I think it’s just a different level of sophistication in the thinking about legislation and regulation. What they’ve said is, for the benefit of society, we want fewer people to die, so we want to be able to do a better job diagnosing different disease patterns. To do that, I don’t need to know your name. I just need to know the information in the medical record, so I can see machine learning similarities.

Sumant Mandal: Right.

Michael Sutcliff: Right? And we’ve said, “Well gosh, we’re not even willing to let that happen. Even if we could, we won’t.” And so, that difference creates a very structural advantage for China in medical research at this point in time.

But you’ll see the same thing with Amazon understanding what videos we like, what music we like, what shopping we like. They’re definitely going to have an advantage over a retailer that doesn’t know that.

Sumant Mandal: Financial services 10 years from now. Consumer finance, where you are, your business, the primary business with Wells Fargo. How does that look 10 years from now?

John Shrewsberry: Sure. It looks a lot more automated. It looks a lot more mobile. There won’t be 10,000, 6,000, in our case, 6,000 branches and 4,000 other pieces of real estate where people interact with us. There will still be some, because people still have complex needs and want to sit down and talk to somebody about it.

But the process of becoming a bank customer, the process of asking for credit, the process of making payments is increasingly becoming embedded close to where those transactions are happening. Very needs based and arising in context. And I assume there will be more and more of that.

Sumant Mandal: And people will always need to go on a vacation. So, how do you see your business 10 years form now, and what are you doing these days to get ready for that?

Michael Flaskey: Yeah. Well, if I’ve done my job correctly and my successor has the opportunity to take this business over, we will transition from a sold product to a sought product. And today, we are a sold product. Today we are a direct marketed and direct sold product. And if I’ve done it correctly, we will have some percentage of our business that is actually sought. Doing business on the web, and big data is the guiding principle to get-

Sumant Mandal: Moving technology.

Michael Flaskey: Absolutely.

Sumant Mandal: That’s great. Well, we have a few minutes left, and if there are questions from the audience, we’d be happy to take them. Please direct them to any one of the panelists, or to the panel.

Audience Member: Thank you. This question is for John. You mentioned that the number of digital interactions between your customers has gone up exponentially. You also suggest that the number of physical interactions have not declined as much, or even at all. I was wondering, have the nature of those conversations or interactions with your physical customers changed dramatically during this?

John Shrewsberry: Sure. It’s a good question. So, digital has gone up, not geometrically, but steeper arithmetically, how about that? And the physical interactions have come down a little bit, but not offsetting.

What’s happening in physical branches, and I think is still much like it was previously, there’s just less of it. In the future that I imagine, there’s less standing in line, making teller deposits or withdrawing cash. There’s really no need to do that today actually, but people still choose to do it.

The interactions in the branch would be more around important life events that have a financial services component to them. College, retirement, house, car, things like that. That somebody, they do it very infrequently, and they need that personal interaction. Of course, that can also be given virtually. But at least at this point in time, people want to do it face to face.

So, in the short term, it hasn’t really changed that much.

Sumant Mandal: Is it any different trajectory wise, once ATMs were introduced versus what’s happening now in the mobile?

John Shrewsberry: So, is the question, did ATM adoption happen at a slower pace than mobile adoption today?

Sumant Mandal: Yeah, yeah.

John Shrewsberry: The thing about ATM is, they had to roll out, which is different than an app on a phone. The sort of handset, smart phone adoption was faster. But the introduction of the ATM sort of forever changed banking. Incidentally now, you don’t need to use a plastic card in an ATM anymore. Your phone plays a role in that, which is an interesting novelty.

Sumant Mandal: Yes. Any other questions? There’s one in the back there.

Audience Member: Yes, hi. I’m wondering who are you seeing in the funds management, investment management industry? Who’s using AI in a very innovative way, and what are the things they’re doing?

Sumant Mandal: Would you like to take that?

Michael Sutcliff: Well, there’s a lot going on in the industry with AI. Obviously, there’s the people like me, investment. So, people are using machine learning to look at patterns of investment for people that have similar profiles, and backgrounds, and risk interest, and they’re saying, “I can recommend for you a portfolio. And I can do auto-balancing much more quickly, much more eloquently.”

The millennials have no interest in sitting down with a 55 to 65 year old investor, and having them to tell them how to invest their money. So, crowdsourcing the information, and letting people understand what others are doing as a basic thing.

On the other side, they’re actually using it to do a much more active and proactive job of scanning for appropriate investments. Very few have actually moved to letting the machine do the final job. WorldQuant, the primary sponsor of this is actually very, very active in that space.

And if you look at all the major hedge funds and all the major investors, they’re all out there trying to figure out where and when the machine learning gives the human a leg up, in terms of making the next best investment decision. So, there’s lots of work going on in this space.

John Shrewsberry: I would add that, in trading execution for all asset managers, I wouldn’t say have any consequence. They’re almost entirely algorithmic in terms of how they accumulate or dispose of a significant position.

Michael Sutcliff: Yep.

Sumant Mandal: And that’s getting in new technologies like Blockchain have a role to play there as well. And I would say that one important sort of thing to take away, at least from this panel is that AI is not as end all be all today for any broad application. It’s a very narrow niche applications where these sort of repeat patterns are being used to learn. And then, decisions made on top of that.

Sumant Mandal: So, we’re still very early, and as I think the panel broadly said, artificial general intelligence is still futuristic, not quite near decision making where humans are.

Any other questions?

Audience Member: I had one. I don’t know if I need the mic. Yeah, one of the things we’ve seen in many industries for the adoption of AI is the lack of understandability. Oftentimes, it can give me the right answer. I’m quite sure it is the right answer. But our human cycles won’t let us accept that. And especially with data coming in where the answer can change not just year to year, but minute to minute. Very hard to compare and see if it’s doing the right thing. Who do you see as leading the way and tackling that as far as industry?

Michael Sutcliff: There’s a consortium was put together on OpenAI, which is working on the concept of explainable AI. And of course, there’s a big difference between machine learning and deep machine learning, in terms of what’s explainable.

The general approach that we’ve seen them take is to basically let the deep learning algorithms occur, and then try and reverse engineer it, and figure out why it came up with the answer. But they’re trying lots of things to try and figure out how to tell humans what it is that the machines learned. And that’s not obvious.

Sumant Mandal: David, do you have a-

David Eun: Yeah, I would say that a lot of the focus of what people refer to as AI, even their own AI has really been on deep learning. Which is using massive scale and resources to analyze massive amounts of data. But there are other emerging approaches to AI, where you won’t have a need for as much data. Where the software itself will become smarter. That the perception of inputs, there’s also a lot of work being done there.

So, what we would see and hear from a machine standpoint, is undergoing lots of change. And then even down to the chip level, there’s new designs for what they call neuromorphic chips. Which are designed not as chips have been, but designed more like how the brain is organized, and might function so.

Again, all up and down the stack there are startups in different companies working on all this to adjust the very issue that you discussed.

Sumant Mandal: Which to one point, auditing deep learning is very difficult because of the massive scale of computation, it’s almost impossible to recreate it and then audit it. So, there are applications where it could be applied, but it’s not being used because they can’t have an audit trail. And I think all of these companies are trying to solve that.

Quantum computing is another step forward, which will at least provide the compute power to go do some of this stuff.

Audience Member: Yes. So, is the science of predictive analytics and AI complimentary, are they one and the same, has one morphed into another? How do you view the two?

Sumant Mandal: John?

John Shrewsberry: I couldn’t draw the distinction between the two. I think of them as sort of advanced analytical capabilities that are being applied to data in different ways today. There may be a technical distinction that you’re-

Michael Sutcliff: Well, yeah, I certainly see them as complimentary, and often working together, but can also work independently. So, both very useful, and absolutely powerful when they’re put together.

Sumant Mandal: Great. I think I’d like to take this option to thank our panel today. Great conversation. Thank you all for coming. And we’re around, if anyone wants to grab them before they leave, please do so. Thank you.

Related Stories