Podcast: Peter Fader, Wharton Professor, Part 1


This is a rare two-part set of free episodes of Marketing BS. My guest today is Peter Fader, professor of marketing at the Wharton School at University of Pennsylvania. Peter was one of my early marketing mentors and I loved this interview. In Part 1 we talk about Peter’s career as a marketing academic and how he came to his signature theories around how one understands the value of a company’s customer base. Tomorrow we will dive deeper into those theories.

You can subscribe to the podcast in your player of choice here: AppleSticherTuneInOvercast , SpotifyPrivate Feed (for premium episodes).


Edward: This is Marketing BS. My guest today is Wharton Professor, Peter Fader. I consider Peter one of my founding mentors for helping me understand how marketing really works. His most important contribution to marketing, in my opinion, is that you can model future customer purchases by assuming that your customer base is made up of a heterogeneous group of customers—each with their own intrinsic purchase rate and churn rate. And that those same models can be used in radically different businesses and industries to create extremely accurate predictions. Most importantly, because these predictions are accurate, it should influence what your actual actions are to grow your business sustainably.

Today, we're going to talk about Peter's career and his intellectual path to this important idea. Tomorrow we'll dive into the idea itself and how it can be used for marketers in practice.

Peter, can you start by talking a little bit about how you first started exploring the idea of Buy ‘Til You Die?

Peter: Sure thing, my pleasure to do so. It's funny because that characterizes my career. That's what I'm most famous for. But (A) it's not my idea, and (B) it didn't even come to me until long after I was a full professor here at Wharton.

I've been building all kinds of different models of customer behavior. How many customers will we acquire, how long will they stay, how many purchases will they make, and all that sort of thing. All the time looking at different data sets, thinking about different business settings, and saying, what would be a story? What would be a model that could capture and then project that kind of behavior?

Back in 2001—again, I had been a professor here for 14 years already—I was building a model to capture a phenomenon that we see all the time. They did a customer-slow-down as they gained tenure with the company. It's pretty universal. I built a bespoke model to capture that and it was good, it was fine. I got the thing published. But along the way, one of their viewers was saying, you want to benchmark your model against this Buy ‘Til You Die model. Something that was invented back in 1987. But it was really technical, it was really obscured, so I thought it was an unfair request.

I went to the editor of the journal and said, don't make me do that. Don't make me benchmarking that old obscure thing. And the editor agreed that I didn't have to. But I wasn't sure he would. I actually did benchmark the models that I was developing against these older ones and found that the old ones were much, much better.

It doesn't show up in that paper. I then decided to devote the rest of my life, or at least the next 18 plus years, to exploring that other model—Buy ’Til You Die. Why it’s so good, different variations of it, different applications for it, different motivations, and different managerial stories around it. That's basically all I've been doing since then. Taking someone else’s model and running with it, calling attention to it, and finding some reasonable success with it.

Edward: When did you realize that it was close to a fundamental law and not something that just might explain some of the data some of the time?

Peter: Because I took it and started applying it to lots of other data sets. Again, this was more out of curiosity than necessity. That's just what we do as scholars which is just try things out. I wasn't only looking at the breadth of applications, I was looking at the robustness even for any one application. The idea that we don't have to have a long data set, and even if we have a shorter and shorter data set, if there is missing data, or if we don't have the same inputs that we get pretty much the same results.

It started convincing me that this is more than just a cute model. It started convincing me that this is actually reality. I know that it's not—and I'm going to lose all credibility with you and your listeners here—but I'd like to make an analogy between this. Brace yourself—the theory of relativity.

We all view that the theory of relativity, E=mc2, and all that stuff, we treat it as if it's true. It's not. It's just a theory. It's just a model. But the thing is it's so robust and explained so many different phenomena, even phenomena that weren't observable 100 years ago when Einstein was putting these ideas out there. But we just keep seeing it “proven” over and over and over again that we just treat it as truth.

Now, I don't want to say that these BTYD models have anywhere near the implications, the importance, the cosmic explanations as relativity. But I think they’re similarly robust and people would just be better off viewing them as if they were true instead of spending so much time pushing back and saying why their situation is different, why the implications don't apply, and why the world is changing. Let's just accept it as truth and our life as managers would be much easier and much more successful.

Edward: But I want to go back a little bit to the path that got you here. I have a theory that things people do when they're 12-14 years old affect them for their entire lives. Where were you passionate about at that age? How did those things affect your later career?

Peter: Oh my goodness. Wow, a bunch of different things, all really nerdy. The one that was most normal would have been baseball. At that time—I'm embarrassed to admit this, you’re getting all this bad stuff out of me, Ed—I was a huge Yankee fan. I've repented since then. I've seen the folly of my ways. I was really, really, really into baseball statistics.

Unfortunately, this was before anyone had heard of Bill James, sabermetrics, or Moneyball. All of that stuff was still years, years later. But I was almost—I don’t want to say—inventing some of those kinds of things but I was thinking very much along those lines. How can we take the game of baseball and break it down into its underlying components, understand those things, and really focus on the underlying story rather than just the overall observable statistics? I was obsessed over that as I still am today.

The other thing is kind of weird. I've always had an obsession with dollar bills with interesting serial numbers. Mom would come back from the grocery store and I would immediately go through her dollar bills. I would say, this one on a 0-100 scale, this one gets a 60. This one, maybe a 40. This one here, that's a 95. I'm going to keep that one. I was just always obsessed with interesting numbers, interesting serial numbers.

Finally, when the whole internet thing started, I bought the domain name coolnumbers.com, and still own it today. That's all that site does is you put in any 8-digit number like a dollar bill serial number and it will tell you on a 0-100 scale how cool it is on my own quirky, arbitrary, don't even try to figure out universal coolness index. It's surprisingly popular. There's a lot of other nerdy people out there, or at least with too much time on their hands. That's the kind of stuff that I was doing. Just looking for patterns in data, but without any particular purpose or societal benefit. I'm really lucky that I finally found some meaningful purpose.

Edward: I'm glad that you're working for good and not evil because I think on the website, you can enter your Social Security numbers. I'm sure people are doing that every day as well.

Peter: Well, right now you can only put on 8-digit numbers. I'm waiting for some kind of undergrad or someone else. Maybe one of your listeners with too much time on their hands to help me flesh out cool numbers. You could deal with, let's say, a Social Security number, a 9-digit zip code, or whatever else. I got the algorithms all worked out. I just need someone to do all the coding.

But thank goodness, I haven't wasted that much more time on it over the last 20 years. I had better things to do.

Edward: You went to college for mathematics, but then you did a Ph.D. in marketing. Why did you switch?

Peter: It wasn’t my choice. There are very few people who say, Mommy, I want to be a marketing professor. It doesn't come up on career day when you're in middle school. It's an interesting story by itself because I indeed was just a solid math major. All I liked doing was crunching numbers, playing around with integrals, and all that sort of stuff. I didn't know what I would do for a living. I figured either end up as an actuary—calculating risks for insurance companies, I’d go to Wall Street, or maybe I'd go work for the NSA and break codes or whatever else.

I was exploring all of these different options until this one professor, this marketing professor, her name is Leigh McAlister. She's still very active today at the University of Texas now, not MIT where I first met her. She came to me one day back in 1982 and said, you ought to be a marketing professor. You ought to get your Ph.D. in marketing. I looked at her and said, you ought to get your head checked because I'm a math guy, I'm not going into marketing. But she laid out this vision—again, keep in mind this was 1982, that's like 500 years ago.

Edward: That’s before finance was even getting into mathematics, let alone marketing.

Peter: But she laid out this picture of what marketing would become. She was exactly right. That there will come a day when we'll be able to tag and track individual customers, know what they're doing, and then get some sense of which message we should send to which customer at which time. We're going to need rock-solid math underneath all that to figure it out, to make these decisions, and to evaluate those decisions.

I didn't believe her, but she was very persuasive and she forced me to get a Ph.D. She literally—I'm not exaggerating—forced me to take this job offer at Wharton. I had offers from lots of other good schools, but she said, “Wharton is the place for you. It will have the people, the resources, the culture to let you pursue your quantitative passions in this domain.”

And here I am. Now, this is year 34 on the faculty, calling her up every 6 months or so, saying thank you, thank you, thank you. She did change my life by pushing me in a direction that, again, I would have never imagined, and even actively resisted at that time. But boy was she right on every one of these dimensions. My whole life is just paying it forward to her in every way possible.

Edward: If you hadn't met her, where do you think you would have ended up?

Peter: Either a Wall Street firm or again maybe an actuarial firm. I took the first bunch of exams that actuaries take. I did an internship with an insurance company. I could see that there was some alignment there, but at the same time, it's not an industry that lends itself to creativity.

I want to come up with new models, new explanations, new stories, just new methods. Whereas in insurance, even on Wall Street, and most of these other domains, it's once you have the way of doing things. It's just shut up and do it. I would have ended up doing one of those kinds of things. Maybe I would have been happy, who knows? I like to make myself happy no matter what's going on.

But nothing could make me happier than the path that I followed. To have the colleagues, the resources, the incentives to come up with new stuff, and then brilliant students, including people like yourself who have taken some of those ideas and run with them, whether in academic directions or in commercial directions. I've just been super lucky to ride their coattails academically and commercially to find success both ways.

Edward: Long before Buy ‘Til You Die, your first significant research was into strategies in a generalized prisoner's dilemma. What exactly did you find?

Peter: Wow. That's a blast from the past. My dissertation at MIT—very few people know this because I tend to focus on all these predictive models of customer behavior and so on. But my dissertation couldn't have been more different.

Indeed, I was looking at the prisoner's dilemma. I'm assuming that many of your listeners are familiar with it already. If not, they can search for it. There's so much out there on it. There's a lot of people who have been trying to “solve the prisoner's dilemma,” coming up with strategies that would be very effective in this very simple two by two game. Do I take the temptation to rat out that person, cut-price, or do the nasty action; or will I be good?

The problem with the basic prisoner's dilemma, as they just implied, is that it has two players—me against you, and only have two alternatives because each of us does the aggressive tactic or the kind of nice tactic. Solving it, in that case, is fine but not very practical because in the real world, there's going to be lots of other complications, and let's just focus on two of them.

Number one, there's going to be multiple players out there. There's going to be three or more firms. In fact, just moving from two to three is a giant leap forward because all of a sudden, if person number three does the nasty thing, what do I do? Do I wait for you—the nice guy, or do I respond to the nasty one? It's very, very complicated and we start getting all confused because if I react to him, then you react to me, and you get into this downward spiral.

Number two, there can be multiple alternatives. Not just do you do the thing or not, but it can be shades of gray. You can be setting prices or discounts or even oil output levels if you think about OPEC.

The generalized prisoner’s dilemma that I put forth had a continuous range of alternatives. It was a price-setting and three players. It generalized, it built upon all the basic ideas of the textbook, two by two prisoner’s dilemma. But it added all kinds of interesting complications, yet it still lent itself to some surprisingly robust strategies. Strategies that I explored in my dissertation. We've seen an interesting range of examples in business, in sports, and in life itself, where some of these strategies do tend to play out and lead to effective outcomes.

Edward: In addition to your research, you've co-founded a few companies. Talk to me about Zodiac and how that happened.

Peter: This goes right back to something I was saying a few minutes ago, which is riding the coattails of brilliant students both in the academic direction as well as the commercial. It’s building out this Buy ‘Til You Die model, and they're really good. They worked really well.

But most of the time, I was either just working on academic stuff to try to come up with new tweaks of them or just going to companies and trying to give them the academic version saying, here you ought to use this. Here, this model is good for you. Here's the code. Here's the spreadsheet. Here's the technical note. Here are some case studies. But the problem is, companies either found it a little bit too academic, or the kinds of data they were looking at was just so messy, so complex, or so large that the academic versions just weren't quite right for them.

Back in late 2014, I had a conversation with one of my brilliant undergraduates. He basically had some ideas to make the models much more practical—to be able to run faster, to be able to run just much more efficiently. Brought in a couple of other folks, and we founded this company. First, we called it CLV Metrics—Customer Lifetime Value Metrics—kind of a lame name. And then we decided, you know what, we're getting such good traction on it. Let's make it real. We brought in some venture capital money. We started hiring a whole team. We changed the name to Zodiac, and it was a wonderful success.

We work with a wide variety of firms. Whether it's retailers, travel and hospitality, telcos, gaming, pharmaceuticals, or lots of different B2B applications and different kinds of services. Just applying this Buy ‘Til You Die model in a wide range of scenarios and finding all kinds of success, all kinds of interesting tactical-use cases—it was really great. But of course, talking in the past tense, because in 2018 one of our clients came along and said, we want it all, and that client was Nike. We sold to Nike in March 2018, which again, was a wonderful outcome by itself, but also a tremendous validation for the usefulness, not just the academic interest in this, but the commercial usefulness of the models.

Edward: We're going to go more into the usefulness of it tomorrow on our second podcast. You later, though, founded another company called Theta Equity Partners and this was different from Zodiac, correct?

Peter: Yes and no. On one hand, there's the no part which is, at the very core, this very similar set of models, this Buy ‘Til You Die model. But the motivation and the main use case couldn't be more different. Back in the Zodiac days, besides working with lots of different companies that I described before, one of our clients was a private equity firm. They weren't that interested in figuring out which message to send to which customer. All they wanted to do was to say, listen, can you come up with the projected value of each and every customer, add all that stuff up, and tell us that number because we're thinking of buying that digitally native women's cosmetics company.

We figured the best way to judge its valuation isn't through the usual top-down multiple approach, but it's from the bottom-up—how many customers will we acquire, how long will they stay, how much will they spend. That's what we did—the idea of customer-based corporate valuation.

After we sold Zodiac to Nike along with one of my Zodiac co-founders, Dan McCarthy, we co-founded Theta Equity Partners. That's all we're doing is customer-based corporate valuation, working with private equity firms, family offices. I'm working with a lot of companies directly just to help them understand, unlock, and fully leverage all of that customer value.

It's less about the marketer. It's just less about the tactics. It's more about finance, valuation, corporate governance, big strategic decisions, and again, it's been great. The models work well. It's probably an even more receptive audience—the finance people than the marketing people. Once you go over the finance people, then it becomes very easy to win over the marketing people as well.

Edward: It's interesting, 38 years or so after you left finance to go into marketing, you're right back where you started with finance.

Peter: I have to admit, I feel like a fish out of water because it's not really my home. It's not my core domain. I've been learning a lot over these last couple of years and I have tremendous respect for the people in finance and more and more every day.

I can bring them a tool that they don't have through these models and through these perspectives. But the ways that they deploy it, some of these are very clever, smart, resourceful things they do, you could see why they are the big dog in most organizations and why people respect, maybe even fear finance much more than they do marketing. Because my objective is to bring them together and to get marketing and finance on the same level using the same models for strategic as well as tactical purposes, and we'll talk more about that.

Edward: Peter,what was the biggest failure point in your career? What's the biggest mistake that you made?

Peter: There's a difference between failure and mistakes. Let me talk about one of each. Maybe the saddest moment in my career—the one night I literally cried myself to sleep—was losing the Napster case. As I've said many times now, I'm interested in a broad variety of applications. I spent a lot of time in the ‘90s and early 2000s working with or maybe fighting with the music industry—there are amazingly good patterns there. It's very predictable. It's one of the better sectors if you want to apply the models, but it's a sector where they don't apply the models.

Long story short, I got caught up in the Napster case, the original Napster, an original file sharing service that changed everything. I was with the good guys. Napster is trying to make the case why that file sharing service is the greatest possible thing for the music industry and making that case why it's good and why it will bring in lots of money.

I wrote this whole long statement, did all this research about it back in the glorious summer of 2000, but Judge Marilyn Hall Patel, she pretty much rejected everything I said. She basically said, the idea that file-sharing could be good for the industry is preposterous and any research that would draw such a conclusion must be gravely flawed. I think those are her exact words.

Edward: Your conclusion was wrong regardless of your methods.

Peter: Exactly right. In the end, it didn't really matter. The reason why Napster was shut down, it had nothing to do with whether it hurt or helped the industry. But the fact is, it was against the law. The law might be stupid, that's a whole other question, so it was shut down. But I took it personally. I felt that this was a true failure on my part. I let down the revolution. It wasn't a mistake. It's just that I was betting on the wrong horse.

Edward: How’d that changed things? Did you change your strategies going forward because of that event?

Peter: Not really. It just made me want to fight harder. It's actually interesting. I said, look, this is just wrong. We need to show the industry that they are making a terrible mistake. In the early 2000s, I spent a lot of time banging on the door of the music industry, saying, listen, let's go after this together. Let's do the research to show the circumstances under which file-sharing helps, hurts, or is neutral. Let's really understand it. Let's understand the business implications. Let's not just stop at music. Let's talk about TV, movies, publishing, and basically all areas of media and entertainment.

I set up a Research Center at Wharton for the Wharton Media & Entertainment Initiative. That went nowhere. Then we got a donation to set up the Wharton Interactive Media Initiative, and that was very successful. That then morphed into the Wharton Customer Analytics Initiative, which continues to flourish today.

I spent a lot of time expanding on it. One might say pivoting from the work in the music industry to try to make a difference with models and understanding of customer data. It's just that the music industry and entertainment, in general, weren’t all that receptive. It's just a matter of shopping these ideas and methods around to find a more receptive audience, which we did find a lot of success with.

Edward: Tell me about the iPhone.

Peter: Yeah, that was a mistake. A little bit of arrogance on my part. I was big into the BlackBerry. I mean that was a transformative device. Wow. When the iPhone came along, I staked out. I went way out on a limb staking out exactly the wrong turf saying, this device will just never catch on. Look at just how different it is. Look at all the features of a BlackBerry that it lacks. I'm never shy about my opinions. Usually, they're based more on data than just pure hunches. This case, pure hunch, wrong hunch, and I basically said that this is going to go down in history as a colossal failure. And again, I wasn't shy about it.

When the iPhone celebrated its 10th anniversary of just a ginormous success a couple of years ago, people went out and found some of these—the incredibly dumb things that I said as it was being launched. I'll admit it. I'm big enough to acknowledge my mistakes. That's far from the only one. But probably the one that I got in—I don't want to say trouble, there's no trouble there—the most shit for and entirely well-deserved. Even though I'm still not a big fan of Apple—I literally have never owned a single Apple device. Again, not that I'm against them but I just like buttons. I like to press things, whatever. I've learned better than to bet against them.

Edward: This has been fantastic. We're going to come back again tomorrow to talk more about Buy ‘Til You Die. Thank you so much.

Peter: Sure thing. It’s always good talking to you.