Monday, April 23, 2018

AI SPECIAL..... The Next Wave: Using AI to Build Better Marketing Campaigns


The Next Wave: Using AI to Build Better Marketing Campaigns

Jordan Bitterman, CMO of IBM's The Weather Company, talks with Wharton's Catharine Hays about marketing in the realm of technology.

As the digital revolution keeps on turning, cutting-edge companies are looking to artificial intelligence to spur the next wave of change. This specialized technology holds great promise for those willing to employ it, but it also comes with challenges. Consumers are familiar with AI in the form of Amazon’s Alexa, Apple’s Siri or IBM’s Watson. Jordan Bitterman is the chief marketing officer for The Weather Company, an IBM business. He recently spoke to Catharine Hays, executive director of the Wharton Future of Advertising Program, about creativity and marketing in the realm of technology for a segment on the Marketing Matters radio show, which airs on SiriusXM channel 111.
An edited transcript of the conversation follows.
Catharine Hays: We like to say that marketers have some of the most interesting backgrounds. Tell us about your journey to get to this role.
Jordan Bitterman: Mine has been a little nontraditional. I spent most of my career at large advertising agencies. That part isn’t necessarily nontraditional, but I did it as a media planner and media strategist. My job was not to create the ads but to determine where the ads go. Most people who follow a CMO track come from account management, which is the team that leads all of the client relationships.
My past was through media, and I started at a time when media was beginning to change in major ways. It was back in the time when a media person simply had to understand an audience, a demographic such as women or men, 18 to 34. But early in my career, understanding audiences started to get far more dynamic and more sophisticated, and that was largely driven by the advent of digital. It became harder to reach people, and it meant that those of us who did media for a living had to become far more strategic ourselves.
Digital media changed everything throughout the advertising industry. I have found that being involved in these leading-edge areas, whether it be web or mobile or social or content, really interested me and helped me along in my career.
Hays: How did you get connected with The Weather Channel?
Bitterman: I joined Weather after the acquisition with IBM. I’ve been here for a little less than a year. I had started thinking about, what does the second half of my career look like? Is it on the agency side or somewhere else? I really zeroed in on the publisher side. But I didn’t want to just work for a publisher who was a content creator solely, I was really drawn to the idea of weather, but more importantly the idea of AI and the Internet of Things. I saw both of those areas incredibly burgeoning. They were areas that I really wanted to focus on. The idea of data as a business, not just content as a business, was something that really appealed to me.
Hays: Maybe you can give us a little background? I’ve been calling it The Weather Channel, but it’s The Weather Company. Why did IBM acquire The Weather Company?
Bitterman: It’s a confusing point that a lot of people mistake from time to time. It requires a very simple explanation, which is that IBM acquired The Weather Company inclusive of everything The Weather Company has with the exception of the cable network that is The Weather Channel. They are now a client of ours. We license them our meteorology, and that’s how they run their cable network. But everything else — the app, the website, all of the data — were purchased by IBM. The main reason for that is that weather is a huge driver of business outcomes.
If you were to look at the earning statements from companies across the board, weather is often a primary reason cited for businesses not hitting their targets for a quarter. IBM, being in the business of helping businesses make better decisions, drive outcomes, obviously saw that as an opportunity. We’re not just forecasts. There’s an underlying platform that fits as a foundation element. IBM is now investing heavily in IOT and cognitive analytics. That was the reason why we were an attractive acquisition for IBM.
Hays: Tremendous amount of data, tremendous impact on businesses, and a tremendous predictive capability as well — those are three really powerful reasons for gaining all of that wealth.
Bitterman: Absolutely.  I’ll give you a couple of statistics. These are things I did not know when I took this job, and they are data points that I have put in front of our clients and customers. We do roughly 25 billion forecast calls per day, and those originate from all sorts of data sources around the world. We do around two to 2.2 billion on an average day of location. Those 25 billion forecast calls are originating, pinging to and from two billion locations a day. It makes us one of the largest IOT platforms in the world.
Hays: In what regard is that IOT?
Bitterman: For instance, we do a lot of business with the airline industry, so the data sources need to come from not just Chicago, Illinois or Philadelphia. They need to come from towers and sensors and pings all over the place in order to be able to realistically have a chance at identifying whether they should put a plane in the air or if they should keep it on the ground. There are sensors inside of the planes. There are sensors throughout the United States. That is the Internet of Things at least in many ways…. When we get down to it, the real mission-critical, bread-and butter elements of IOT are what is sitting in elevators and airplanes and turbines.
Hays: Watson is The Weather Company’s new best friend. Give us a brief history of Watson, which I think really has been this country’s first interaction with artificial intelligence.
Bitterman: It’s almost a foregone conclusion that someone who talks about Watson will start with talking about “Jeopardy!” It started as a way of demonstrating whether it could work or not, and almost internally than externally. It started as one natural language, QA, question and answer, application program interface (API). Today, it is represented by a whole diverse set of Watson’s services that span everything from language to speech to vision, etc.
You may recall that Watson went on “Jeopardy!” and won. It was the first time that we, as a technical society, really started thinking in many ways that computing can get into more than just binary. Watson is best understood by thinking about it through the three eras of computing. The first was the tabulating era. Think about Imitation Game, the movie about Alan Turing. It was essentially calculations. That’s practically up to the 1950s. Punch cards, etc.
From there we went into the programming era, and that was about using computers to help solve equations where we know what the answers are, or we know generally where the answers are going to lie, but we want to get there faster and more easily. When I think about the programing era, I think about Excel. But you could pretty much think about any other piece of software that we’ve used over the decades.
But now we’ve entered the cognitive era. What separates the tabulating era and the programming era from the cognitive era is that we can use structured and unstructured data. It’s not just zeroes and ones. We can lean into unstructured data like video or audio or faces, expressions. The idea that you and I could be sitting across a table from each other and I can read and react to your expressions based on whether you think I’m saying something smart or something kind of off. Computers can start to do that now, and that’s really what Watson is trying to do. It’s trying to learn, understand, reason and interact back. A definition of cognitive that I particularly like is: Programming is where people program computers, but cognitive is where computers program themselves. Certainly, with the aid of people, but they can start to program themselves, and that’s when things get super interesting.
Hays: That’s a really good distinction. Tell us about how Watson has pervaded IBM. Help us to navigate how IBM is thinking and organizing around it.
Bitterman: Obviously, IBM is a very big company with many divisions that solve problems and challenges for clients across the world. Watson is two things. It’s divisions within IBM, and it’s also offerings that help drive some of the other divisions within IBM. The Watson content and IOT platform is part of a broader Watson ecosystem, and that broader ecosystem addresses health care, financial services, retail, education, legal. There is development that goes on at Watson. There are teams of engineers and technologists who are building out more aspects of the program. It launched in one language, but it’s being trained in eight additional languages now including German and Korean. It’s a diverse set of services that span language and speech.
For our group specifically, we’re made up of a few different parts of the business. We’re made up of an IOT platform. I liken it a lot to the Romans and the aqueducts. There was water. We knew we needed water. We had to get water from one place to another because without it you couldn’t irrigate your fields, you couldn’t drink the water, you couldn’t make wine. Once you built the aqueducts, all of a sudden water was set free and could be used in so many places. That’s what we’re doing with the IOT platform as it relates to data. How can we take data and make it more useful? Certainly, there’s data that is useful all over the place right now; we’re not really stretching for data in this world. But there’s so much more data, like the data I’ve talked about before that come from elevators and airplanes, that we’re not using enough of yet. That IOT platform is charged with that.
A large part of our business is The Weather Company and being able to build solutions that might start with weather but don’t end with weather. How do we enrich it with cognitive technology, and how do we put that to use for our clients?
Hays: Watson also is getting into the business of advertising. As if agencies and advertisers don’t have enough threats, here comes artificial intelligence. Tell us how AI has impacted advertising generally and what role IBM and Watson are playing in that field.
Bitterman: At IBM when we talk about AI, we don’t actually use the term artificial intelligence. We use the term augmented intelligence. It might sound like marketing spin, but words matter. When we think about augmented intelligence, we don’t think about any of the solutions that we have putting agencies or putting anyone out of business. We actually see it as benefitting them. If we go back to the programming era, computers didn’t put advertising agencies out of business. Computers spawned an entirely new way to do advertising, both to display advertising but also to go about creating media plans, etc.
We think about how AI needs to be put to use in the advertising world. We started out with something we called Watsons Ads, and it’s a creative tool in a lot of ways. You could think of it as one of the world’s coolest rich media units, and you probably know rich media as just a more robust kind of creative technology. When we launched Watson Ads just about a year ago, we did so with the idea that much like Watson itself, which is all about learning and understanding and reasoning and interacting, our ads would be able to do the exact same thing. Picture an ad wherever you are, on a mobile device or on a website. We’ve all seen the ads, and a lot of times we don’t want to interact with the ads because we feel like they aren’t helpful to our session on our mobile devices. They’re actually hurtful. They’re interruptive. But in this case, we want them to be conversational. We want to bring value to people. For instance, with Campbell’s, you are able to tell the ad what ingredients you have in your home, and the Watson ad will give you a recipe that you can make with those ingredients. What makes it cognitive is that the ad can’t possibly know all the combinations that you’re going to ask. It doesn’t know that you have leeks, curry and chicken. But it’s cognitive, so it learns the kinds of things that are great combinations to put together. It works by giving you a helpful result, and it gets smarter as it learns.
Hays: Is that voice activated?
Bitterman: Yeah, it doesn’t have to be. You can type into it. But one of the really interesting aspects of cognitive technologies is that it does have natural language recognition skills, so a lot of the clients that we’ve worked with on Watson Ads have taken advantage of that API. Because why not, why wouldn’t you want to give people that option?
Hays: Especially if you’re starting to cook. Your hands are dirty and you don’t want to type. Voice interface makes a ton of sense in many applications.
Bitterman: It does. We see that with Amazon with the Echo, and we see it with Google with Home. Those are very much consumer applications. IBM and Watson are not in the consumer business. The Weather Channel does have consumer apps, but for the most part when we put Watson to work, even on our own properties, we’re doing it with the idea that it’s a solution for enterprise that we will take advantage of. In this case for Watson Ads, it’s a solution that Campbell’s or GlaxoSmithKline or whomever can take advantage of to interact with their customers.
Hays: You’re redefining what an advertisement is, which is part of what we talk about all the time and why we called our book Beyond Advertising, because it is creating value. It’s not the typical ad. You’re not saying that it’s trying to come up with whether it’s red or pink or yellow, whether it’s a bull or a chicken or something else. You’re saying that it’s this interface, this interaction between the audience and the brand where something good is happening for both.
Bitterman: We are not shying away. In fact, we are going to have more products, more cognitive advertiser offerings that we’re going to be launching in just about a month. But we started out with one that we thought Watson Ads, the creative that I just described, would do a really good job of emotionally connecting not only with the audiences that our clients are trying to reach, but also our clients themselves. For them to understand this is where the future of advertising is going. Watson Ads has been able to telegraph that in a very meaningful way.
Hays: Are you finding advertisers generally hesitant or excited about AI?
Bitterman: It varies, which I know is the worst answer to ever get. It depends. The most innovative brands are leaning in very hard on Watson Ads because they see it as an opportunity in exploring new technologies and using our technology to engage audiences in new and breakthrough ways. But building a Watson ad takes a little bit of time and a very considered approach. If you have a marketing team of two and a lot of things that you’re trying to accomplish, then you’ve just got to stay real focused on getting those things done.
But if you’re a bigger brand — Campbell’s, GlaxoSmithKline, Toyota and Unilever have all worked with us — they have higher risk thresholds because they are open to trying things and experimenting. I don’t mean to suggest that what we’re doing won’t work, but it certainly takes a leap of faith. It’s not the first thing on your plan. Just like me as a marketer, the first thing on my plan is the acquisition marketing. What is going to drive revenue or drive traffic to our consumer properties? I need to have bigger budgets and a bigger risk threshold to try to ride the crest of the wave and do something new and different and move into the future of advertising. The bigger clients understand that.
Hays: It’s almost riskier at this point not to start jumping into the game, not to start understanding how it can be relevant for your business and how it can enhance what you’re doing from a marketing perspective. Do you think it’s too early for some companies, or do you think everybody should at least try getting started in this field?
Bitterman: I’ve been in our business a long time, and I’ve got the gray hairs to prove it, and I’ve had the good fortune to have lived through the digital revolution and the mobile revolution and the social revolution. If you think about all of those parts of our industry, those times of our industry where we have seen so much change, we have arrived at a place now where all three of those areas — digital, mobile, social — are all mature industries.
I do remember working with colleagues and with clients when I was on the agency side who were either indifferent to those areas or were sort of scared of those areas. Those areas have all become big businesses with anywhere from two to five billion people on the planet consumed by it. But if you think about AI, AI is going to consume the total population of the planet. The reason why that’s true is because you don’t necessarily opt into AI the way you opt into social when you create a Twitter account or a Facebook account. You either opt into AI because you’re wearing a Fitbit or you’re giving a company permission to use the data that’s coming out of your dishwasher or it’s working with you in an ambient way. It’s changing the traffic light grid in your city. Whether we know it or not, whether we opt in or not, AI is touching all of us and will touch every person on the planet.
Hays: It’s what’s powering some of the most interesting technologies. As you were saying before, you can’t have IOT without augmented intelligence. In that regard, it truly is becoming pervasive. So how about the risks, especially from an ambient perspective, something that you’re not opting in but it is checking your biometrics as you’re walking past? What are the risks that you are seeing and the responsibility that a company like IBM has?
Bitterman: Our company has an ethical and moral standard that is unlike I’ve ever seen before. Some days in my job it actually frustrates me a bit because the bar is so high in terms of what we can put out and what we can’t. But even though the day to day can be frustrating on that particular topic sometimes, it’s really best that we have big companies that think that way.
Back during Mobile World Congress in the spring, the CEO of SoftBank had a slide up in his presentation, and he said that there were 12 threats to civilization. One of them was AI. But he said the only threat among the 12 that actually is an antidote to the other 11 is AI. AI can help solve those challenges. We do have to regulate, and I believe that the government is behind in even thinking about this. Companies at this point are self-regulating, and that’s a good thing. We’ve got a lot of companies in our world that have great intentions and they’re on it. They might not be on it all the way, but I know I can speak for IBM when I can say that we are. But over time, we as a society have to think about these things. What are the challenges that we want to solve for? And let’s make sure that while we’re solving for some of those challenges we’re not creating new challenges in the process.
Hays: Could you tell us about the Toyota ad campaign and what you learned from it as we’re at these early stages of augmented intelligence.
Bitterman: Sure. You can ask the ad pretty much any question you want about the Prius. For instance, I wondered about the charging process. I live in New York City and my garage doesn’t necessarily have access to special charging docks. I asked it a series of questions in my own voice as to how I need to charge, what kind of outlets I need to use, how long does a charge take, how long does it last? All those answers came back, similar to when we were talking about Campbell’s and the ingredients. But they get you to a point where you feel more connected to the brand, to the specific model. As a marketer, I know my job is to diminish the distance that exists between an audience member, a customer and a sale. If I can find the way to diminish that distance, to make that point A to B shorter, in this case get people to come in and take a test drive, then I’ve done my job. That’s what Toyota Prius and their Watson ad is all about.
I can’t speak to Toyota specifically because that would be their right, it’s their data. But I can tell you in aggregate when we look at the various Watson ads that we’ve created, a number of things have happened. No. 1, we’re seeing one to two minutes of active engagement with the ads on average. Some people spend a lot more time with it. But one to two minutes with an ad is a lot of time. That exceeds digital industry time-spent benchmarks.
We’re also revealing lots of interesting product insights, like consumers tend to engage more deeply when they’re submitting what we might call non-intuitive ingredients for a recipe suggestion, or when they’re asking questions that are outside of the box of what someone might normally engage in with an ad for a car manufacturer. Not only do all those things they help an advertiser understand how the ad is working, but it also could create lots of insights for how they might want to revamp their product brochure or their social strategy. Because these are all tells that couldn’t be had in any other way.
Hays: Speaking of humanizing these sorts of things, if the Watson ad gets flummoxed and you have a question that it doesn’t know how to answer, is there a human interface?
Bitterman: It’s all in how you architect the ad itself, but the ad keeps working. It keeps trying to answer your question. It will ask you to ask it in a different way. It will give you a suggestion. When we architect the ads, we want to make sure that we’re deep-linking people into a very specific experience based on what we’re hearing about them. You can think that the technology that’s in a Watson ad could absolutely take someone deeper, that you are basically closing the sale or taking someone further from engagement into consideration, or from consideration into purchase. The goal is, how do we get people and diminish that distance between where they are now and where they need to be?
Hays: What do you see as the top three trends that we should be looking out for in this space?
Bitterman: You mentioned before the issues around privacy and security. All of us in this industry need to be thinking about that because if we don’t, it’s going to be done for us. It will probably be done for us anyway, but we should be building an ecosystem that is really right for the long term, for what we want to build, not just for where we are right now.
I think we need to focus. Focusing is very important. Right now, AI is very large and kind of like digital in itself. Back in the day someone would ask, how much does a website cost? How much time does it take to build a website? This goes back to the early to mid-1990s. Right now, some of those same very broad questions are being asked to AI. Do I need AI? How much does an AI ad cost? The goal here has to be to focus. We have to be able to focus both as a provider of AI services, but also as marketers who are out there trying to engage with us. We have to both sell more focus and we have to buy more focus in order to get to where we need to go
http://knowledge.wharton.upenn.edu/article/ibms-watson-future-ai/?utm_source=kw_newsletter&utm_medium=email&utm_campaign=2017-08-10

No comments: