Episode: 452 |
David Edelman:
David Edelman on AI in Marketing:


David Edelman

David Edelman on AI in Marketing

Show Notes

David Edelman spent 30 years as a Chief Marketing Officer (at Aetna/CVS), and consultant to executives on Digital and Marketing Transformation with McKinsey, Digitas, and the Boston Consulting Group. He is now a senior lecturer at Harvard Business School and serves as an advisor to top executives in start-ups, Private Equity, and larger enterprises. In today’s episode, David talks about the use of AI in marketing. You can reach David through his LinkedIn profile or email him at dedelman@hbs.edu.

Key points include:

  • 03:44: The different aspects of marketing
  • 08:38: How AI can be used in marketing
  • 15:50: The prerequisites a company needs to employ AI
  • 21:16: Measuring the ROI on AI-driven tools

One weekly email with bonus materials and summaries of each new episode:


David Edelman


Will Bachman 00:01

Hello and welcome to Unleashed the show that explores how to thrive as an independent professional. I’m your host will Bachman and I’m so excited to be here today with Dave Edelman, who is a professor at Harvard Business School is the former chief marketing officer at Aetna. He was a partner at McKinsey, where he helped build the digital marketing practice he helped found digitize he started his career at BCG. Dave, welcome to the show.


David Edelman 00:28

Thanks. Well, it’s great to be here.


Will Bachman 00:31

All right, I am. So thrilled that you’re on the show today. I think we’re going to talk about an area where you’re doing some work on of AI, artificial intelligence and marketing. But before we get to that, I’ve got a question for you. Really naive question, which is, what is marketing?


David Edelman 00:54

You know, it’s funny that you asked that well, because I think from the conversations I’m having with CMOS, and from my own experience, marketing is going through an existential look at itself to figure out how it should ironically, brand marketing, at its core, what marketing really should be, is, has got two sides. One is helping companies think through all the strategic issues they need to address to build a customer base. So what does that mean from the kinds of insights they should have about their customers, segments that they should focus on serving the right offerings for those segments, the best ways to reach those segments, and then to continue to build relationships with those segments. But then you also have to execute, and you have to actually change behavior. And marketing is not only about the strategies, but it is also about the tactics, that actually change customer behavior, in order for a company to achieve those objectives that it sets in its strategy. So there’s two sides to marketing, there’s the strategic and the tactical, I think, in many cases, people just see the tactics, because those tend to be the more obvious ones in the market. It’s also where a lot of marketers have grown up from their junior days or starting at agencies that focus with tactics. But for companies to really get the value of marketing, they should be thinking much more strategically, and how marketers can actually lead the company in driving a customer insight based strategy.


Will Bachman 02:50

Very helpful. You sometimes, you know, some people might think of marketing as, as advertising or as like the branding, and the logos and so forth. And then some people will say, like, marketing is everything a company does, it’s how you answer the phone, it’s, you know, what employees believe and tell their neighbors about what they do at work at, you know, and so there’s like this very broad kind of range of to what degree we think about it with within that tactical and strategic aspects of marketing that you mentioned, like, what are the different sort of functions or pieces of marketing? Can the layer below the CMO or two layers below? Like, what are the different activities that go on make up the marketing department?


David Edelman 03:44

Sure, marketing does have many different aspects. And again, it depends on how the company defined some of the borderlines. But in a robust marketing organization, you start with market with strategy, where you’ve got much more strategically oriented folks who are working with the business unit leaders, to figure out the best ways to go to market launch a new product, focus investment in a new offering that’s going to extend the lifetime value of a particular segment. These are marketers who are embedded with the product teams, and who are bringing insights and an understanding of the art of the possible to the table so that the product the US can really understand how to grow the business. alongside those going to be brand management brand is absolutely critical to every company, and marketing has to be the ones to bring out what the brand can be and how to make sure that brand is consistently accurate. cuted in generating value across the company. So what should the brand positioning be given what the company’s mission is? How to translate that brand into all the stuff, you obviously see the messages, the creative, the tone, the font, the visuals, all of that, certainly marketing manages those aspects of the brand architecture. But it does go down to many aspects of the customer experience and the employee experience, how people feel their interactions with the company, add value, what is the sensation in the emotion, besides the actual rational things that happen? What’s the emotion that you get from interacting with a company, whether it’s in a call center, whether you see an ad, whether you buy a product, all of that is really the brand, and the brand, and the way it wants to communicate should permeate all of that. And that’s one of the important reasons why marketing needs to be at the leadership table, then there’s a whole lot of execution expertise centers of excellence, that help actually make marketing hit the ground. So you’ll have marketing operations about getting campaigns and programs out the door, many of which may be worked on in association with an agency, you’ll have media analytics media strategy, again, the media buying, they go to an agency, you’ll have different degrees of creative capabilities in a company whether inside or outside. at Aetna, we built our own in house creative agency of 100 people. Because we especially felt as we launched a new brand, we wanted to have a lot more control over how that brand manifested itself, not just in advertising, but in the over 3000 different types of messages that we send to our members every year. So there was just a vast range of touch points that we had to infuse the brand into. You’ll also have marketing analytics, which is becoming more and more crucial, as you think about how to segment how to target how to set up tests, how to use those tests, to then think about allocating money. And then of course, marketing technology. Think about all the platforms that are going to be essential to help you reach customers to manage interactions to enable test and learn all of that, from a marketing standpoint about what you got to do to drive your growth goals. So it’s a pretty broad range of functions, ranging from the strategic business partners, to the more creative aspects of the brand and the tactical of getting stuff out the door. The much more or less brain aspects of analytics, and the foundations of technology, all of which could be my team was pretty broad ranging in its skill base.


Will Bachman 08:16

Wow, I just want to frame that that whole discussion. This was this was so helpful as a one on one for me. I haven’t I haven’t heard it so succinctly before this is great. Let’s turn to AI in marketing, you’re been doing some thinking there. Tell me tell me your thoughts about what’s what’s happening with AI, how can be used?


David Edelman 08:38

Yeah, I’ve been doing work with a range of startup and midsize companies who are bringing very real capabilities that are going to transform aspects of marketing. And I’ll touch on three quick examples. One of the most important things in marketing, increasingly is the ability to connect with a customer on a personalized level. So you’re sending an email, you’re sending a text message, you’re rendering what should happen on a website, in order to match what you know about the customer where they are in their journey. And so you kind of know what you want the customer to do and what the customer expects, but you don’t necessarily know what’s the exact right. Creative message, should there be an incentive? What’s the timing of the message? There’s an endless range of variables that you got to think about when you personalize an interaction. And the traditional way marketing does this is through a lot of A B test and learn split tests where you try champion and challenger. But the problem is that really doesn’t scale across all the different variables and all the possibilities of personalization And we learned this firsthand at Aetna, when we were trying to launch programs to get people to take healthier actions. There’s just an endless range of variables. And so you start to bring in more advanced techniques like multivariate testing. But even then, the scale of what you want to manage complexity is really high. So for example, one company I’m working with, they’re called offer fit. And they bring AI into multivariate testing. So they use computer machine learning models, to constantly throw out and create automatically new kinds of tests against different segments, see how customers respond, capture the learning from that, and then continue to either test new things, or use what it learns to optimize existing interactions. And the results are just incredible. It’s growing like mad, I think it lessens the need for companies to have as many data scientists as they’re trying to hire right now, you’ll always need data science. And that’s not at all. But I don’t think most companies can get the volume that they think they’re going to need to run personalization programs. And you’re going to need something like artificial intelligence tools, like offer fit to be able to make that work. The second example, is understanding where customers are in their journey so that when you do message them, you have a sense of their context. And it’s been very hard to understand all the different touch points someone has had and how to stitch that together in a time series to actually follow a journey. records are kept in many different computer systems, hard to make matches. And so now there’s AI capabilities that can do identity resolution, so that you can stitch together will Bachman touchpoints, all across Comcast, including the usage of the cable box service issues, paying bills, and spot where there’s problems, or where anomalies are happening that you’ve got to address. Are there specific things right now where things are going wrong, such as you called the call center three minutes after using the app, that sounds like you probably had a problem on the app. So we got to address that. So these kinds of analysis are now possible through customer journey analytics tools. One I’m working with is called pointillist. There are others out there. But they’re bringing new insight into using two kinds of AI one for identity resolution. The second is for spotting anomalies, and bringing those to the surface so you can act on them. And then the third example is video, and personalizing video, where right now, a lot of cost goes into creating videos to enhance sales in advertising various aspects of the customer experience. But what if I can personalize a video to you. And when you become a new member of Aetna, for example. And we did this, I can send you a personalized video that tells you will Bachman exactly the plan you bought all the economics of it, what what capabilities of your plan you took advantage of last year what you didn’t tells you whether or not you’ve got a primary care physician if you don’t hear a once in your area to just click on and sign up. And so it goes through how you can take most advantage of the plan economically and health wise, and company called Sunday sky we worked with to do personalized videos 70% of our members who were sent these videos watched a four minute video, the response was amazing. The reduction in emergency room visits was notable because we help people understand the constant unnecessary Enos of going to an emergency room for a lot of routine things, for instance. So there’s real impact here. And as you think about all the different ways AI can influence the way a customer experiences you the content, your ability to spot anomalies. I really think more and more of the way companies compete and the way their brands are manifested in a customer experience will depend on how they’re using AI to bring better information to bear in every interaction.


Will Bachman 14:58

Talk to me a bit about The capabilities that a company needs to have in place to start making use of some of these tools. And I imagine it’s one of these kind of, you know, consulting firms, Gartner McKinsey, they always like to have these different, you know, staircase levels like level zero is you’re completely unable to do this, because you can’t even just do it, you know, using standard methods. And number four is probably the, you know, everything is AI to the nth degree. But what sort of capabilities does a company need, should a company like try to first get to a certain set of prerequisites before they think about using these capabilities like, don’t do it, if you can’t even do a B testing yet? Talk to me about sort of, you know, who should start thinking about this and, and on that ladder?


David Edelman 15:50

Sure, it isn’t something that you can just throw on to accompany just like this, there’s two major prerequisites that you absolutely have to have before you start really dancing with AI. The first is a sense of how you’re going to personalize things, a big part of AI is taking things down to an incredibly granular level. So before I talked about offer fit, in terms of the contents of the sky, in terms of video, you’ve got to already be thinking about your interactions with customers are going to need to be personalized, and that you believe in the value of doing so. And you’re starting to think about what information you have that can allow you to personalize those interactions. And that you’re starting to build an arsenal of knowledge that can be used as fuel, to drive the decisions on how to personalize. So that’s one. The second is you need to be thinking about marketing, as not just advertising and such, but as the broader customer experience. So thinking about the broader journey that somebody is going through with you, because there are going to be many different touch points that you’re going to want to personalize. And you’re going to want to apply technologies like AI to help you understand help you understand where’s the best touch points to invest in in the first place, and how to make those touch points better. If you’re just thinking about marketing as just very basic, managing an acquisition funnel, yes, there’s AI capability that might help you choose your media, those tools already exists in what’s already inside of Salesforce Einstein or Google Analytics, that those exists there. They’re already there. But if you go into really build AI much more into the fabric of how you want customers to experience you, you’ve got to be thinking about personalization, and experience. And starting to collect data and recognize that data actually has real important strategic value to change the nature of the experience you’re delivering.


Will Bachman 18:25

Do you have any thumb rules around the scale a company needs to be at or it makes sense to think about this, it seems like with the personalization, for example, you need to have a certain volume of, you know, customer interactions on your websites or visits to be, you know, to be wanting to move beyond simple A B testing to a much more sophisticated AI driven testing. So is there like a number of customers or a number of visits or $1 value of revenue or some kind of metrics that you think about, hey, if you’re below this eyes, really just like don’t worry about it?


David Edelman 19:04

Yeah, I don’t think it’s a revenue number per se, because it’s about the nature of interactions. And it depends on what you’re actually trying to optimize. But generally, you need over a million interactions to start with, and almost all the platforms that I work with, but the baseline is you got to have a million interactions with some frequency on let’s say, a monthly basis to be able to feed a machine that can then start making judgments about differentiation across a population. So you need to you need to get to that kind of scale, but a lot of businesses are already there. And I’ve been surprised at how even smaller, midsize and smaller companies are using this. Just give you A good example of a company I’ve come to know well brings home security. They’re not the biggest player in the market, they’ve got about 2 million customers. You know, ADT, and others are much bigger than them. But they have gone whole hog into AI to use this to optimize all of their retention offers, think about scheduling of maintenance, how to manage customers for upgrades, everything about customer interaction, and the decisions that they make across their customer base. They are moving to some kind of artificial intelligence engine, and are seeing really impressive results. Bill Nye is the CEO, they’re amazing guy who really preaches the gospel on this brought the company into a much more leaning forward attitude about all the data that they have. And it’s really making a difference in the way he’s transforming them.


Will Bachman 21:00

How should a company think about measuring the return on investment of moving to this AI driven tools? What sorts of investments are required? And? And yeah, and how do you think about ROI?


David Edelman 21:16

Yeah, I mean, besides the tools themselves, which are priced in all kinds of different ways, from, you know, some are priced on share gains, some are priced on just simply usage levels. So there’s different economic models depending on the company you’re working with, and the nature of the tools. But there are some other investments, there’s investments in many cases that you’ve got to do, in your basic data, cleaning up some of the core data structures that you’ve got, setting up the processes and people to just continuously manage your data hygiene and manage it as a strategic resource. So those are incremental things, you do have to consider it an investment, in terms of thinking about the ROI all comes down to the specific use cases. And you have to, before putting in an AI system, you don’t just simply put in an AI and say I’m going to do is you think about very specific use cases that you want to drive. Because the AI is trained against those use cases, it can’t do everything. You got to gradually train it to take on certain use cases. And then you keep adding more use cases into the mix. So if your use case, for example, is digital containment, is a pretty common one where you want to keep people online not calling in the call center in order to save money. That’s a good example. So you want to that’s something where pointillist, for example, that looks at the customer journey and where there’s problems. One of their most frequent use cases is digital containment. And so you look at the decreasing calls to the call center that are happening and the rise and completion of digital interactions. And you can see the cost savings from reduced call volume. And presumably you might also see better usage satisfaction, other outcomes from better use of the digital tools, more engagement could drive more usage of your product. So it really it’s very use case specific. And it’s important to understand those right from the start.


Will Bachman 23:39

Fantastic. I want to ask you a turn topics and I want to ask you a little bit about your experience as a professor at Harvard Business School. I think a lot of executives might get a dream about doing that about, you know, semi retiring, but not I mean, from a day job as an executive and big into becoming a professor at business school. Love to hear you. What have you learned so far in this either from preparing your course, or from students? And what surprised you about the experience?


David Edelman 24:15

Yeah, it’s been a wonderful change of pace. I think a few things that I’ve learned. One is that teaching is a new skill. Just like you know, any functional job has a range of new skills, the actual act of teaching. And at Harvard, they use the case method. So it’s a performance art. I mean, you are in the pits physically, working the class, getting them to jump in creating opportunities for them to debate each other. You’re constantly challenging them. I get pretty sweaty, in the courts of a classroom. It’s very physical. So that’s one thing that surprised me. I think the second thing is just my own humility. As I go back through learning, marketing from the lens of cases and all the just a lot of basic stuff that because I just wasn’t working, let’s say in packaged goods for a while just relearning different dimensions of what’s important for marketing in those areas. Plus, one of the great things that’s happening in most business schools is the globalization, the international student base, my students, 40% of them come from outside the United States. And more and more the cases are international to with really fundamental differences in things like what’s available online, the nature of distribution channels, cultural issues that affect things. And so I’m learning a ton. And really expanding my mind about all the different areas, that marketing makes a difference and how to bring a consumer lens and with very different perspectives than I myself have worked in before. So it’s really been fun. The students bring in a lot of new angles that I hadn’t thought of yet. And along the way, I’m also building a course. Eventually, this I’m teaching first year marketing now I’ll be building a course on competing on customer experiences going forward. And so in the act of doing that, there’ll be I’m sure, plenty of learning as well.


Will Bachman 26:32

Fantastic. David, if people wanted to follow up, and, you know, find out what you’re working on and what you’re doing, where would where would you point them online?


David Edelman 26:44

Sure, a couple of places. I mean, one is certainly my LinkedIn page. So I’ve got a pretty major presence on LinkedIn, I publish a fair bit there, I have to admit, I do have over a million followers that I built up some blog posts on LinkedIn over the years. So that’s, that’s one great way. And you can also always email me and reach me at this point through the business school. So you know, I have a business school email address, D edelman@hbs.edu. Got a profile there. And you can reach me through that channel as well.


Will Bachman 27:22

Fantastic. Well, David, thank you so much for joining today’s it’s fascinating to get a tour of AI and what’s happening there and marketing and get a basic definition of the of the function. It’s great speaking with you.


David Edelman 27:37

Thanks. Well, it’s been a pleasure as always

Related Episodes


Nicolai Chen Nielsen, Advisor, Author, & Entrepreneur

Nicolai Chen Nielsen, Advisor, Author, & Entrepreneur


Tom Critchlow, Writer and Strategy Consultant

Tom Critchlow


Author of The 2-Hour Cocktail Party

Nick Gray


President and COO of Bunkerlabs

Joe "Hark" Herold