August 15, 2016
Interviewed by: David Snow
Video Clip
Login to view full video

Wharton’s Fader: How to Use Big Data to ‘Value’ Customers

Private equity firms that invest in the consumer space are paying close attention to the research of Peter Fader, a professor of marketing at Wharton who tries to predict the value of a company by predicting the life-time value of its customers. In a Privcap interview, Prof. Fader says that in too many cases consumer and retail companies that don’t fully understand their customer data can’t optimize their operations and undervalue their franchise.

Prof. Fader will be presenting at Privcap Game Change: Consumer & Retail, Oct. 19 in NYC. Learn more here.

Private equity firms that invest in the consumer space are paying close attention to the research of Peter Fader, a professor of marketing at Wharton who tries to predict the value of a company by predicting the life-time value of its customers. In a Privcap interview, Prof. Fader says that in too many cases consumer and retail companies that don’t fully understand their customer data can’t optimize their operations and undervalue their franchise.

Prof. Fader will be presenting at Privcap Game Change: Consumer & Retail, Oct. 19 in NYC. Learn more here.

Using Big Data to Value Customers
With Prof. Peter Fader of Wharton School

David Snow, Privcap:
We’re joined today by Pete Fader of the Wharton School of the University of Pennsylvania. Dr. Fader, thanks for being here today.

Peter Fader, The Wharton School of the University of Pennsylvania:
It’s wonderful to be here with you, David.

Snow: You are an expert at consumer behavior. You’ve done a lot of research into how consumers behave, how they can be tracked and how you can predict their behavior using data. Before we get to some of the breakthroughs you’ve made in your research, I’d love to get your perspective on what the traditional challenges have been for retail and consumer-facing companies in trying to track what their consumers are doing and to predict what their customers are going to do.

Fader: The biggest challenge is ignorance. If you look in a dictionary under the word “data scientist,” you’ll get all kinds of definitions for it. But if you find the antonym—the opposite—of a data scientist, it’s a retailer. It really is an industry that prides itself on how our CEO started in the mail room and worked his way up and even though it’s a wonderful opportunity to be using data, leveraging it, getting smarter and making more money as a result, it’s an industry that—because of very strong traditions—has not only avoided it, but has in some ways shunned it. Of course, there are some exceptions, but the rule is, I can tell more about a customer just by looking at the way she walks in the store than I can from any of that big-data stuff you can give me. So, there are problems, but there are opportunities.

Snow: Talk about the work you’ve done and maybe fast forward to some of your most recent findings.

Fader: I’ve been on the faculty at Wharton for 30 years now. I spent all that time predicting things—predicting who’s going to come back when and how often and what they are going to spend and what they are going to do after that. Where a lot of it all comes together would be this idea of customer lifetime value. If you can give me a bit of data and do a bit of razzledazzle math, I can give you a pretty good guess about what any one customer is going to be worth in the future. In some sense, [it’s] a simple discounted cash flow, but it’s not so simple when we’re talking about customers—human beings who can be very random and very different from each other.

Our forefathers in direct marketing—which, again, is a very different sector from retailing but shares a lot in common when you strip away the industry practices and look at the data. Our forefathers gave us the rubric of RFM. They told us that if you want to predict things like customer lifetime value, it hinges upon recency, frequency and monetary value—very simple things that you get out of a very ordinary transaction log. A lot of the other stuff that companies (not just retailers, but all companies) are chasing after these days—like where someone’s located in the social network or what kinds of social media postings they’re putting up there, or what’s going on inside their brain—a lot of that stuff is nice to know. But, if you want to make really good predictions, it’s amazing how such simple, well-established metrics (RFM) can give you really good predictions. Again, you can layer on more or get more complicated, but…in many cases, it hurts the forecast if you overcomplicate it.

Snow: Can you give some examples? Without maybe necessarily naming names of companies, can you give some examples of how a company leveraged the data that it collects on customers to optimize its operations?

Fader: I want to name names and I don’t think these companies would mind. In fact, I’m out there looking for companies that are doing this stuff particularly well, often in the retail setting, and being exemplars for everybody else. I’ll give you two quick ones that…revolve around this idea of understanding the future value of the customer. One of them would be Electronic Arts, the gaming company.

They’re amazing because here’s a company that was born around the idea of coming up with blockbusters. Let’s come up with a big, great game that everybody’s going to buy. That world is saturated, plateaued and very commoditized. So, the real breakthrough for them is understanding the value, the future value, of different customers who are buying the games.

There’s a specific example. They are looking at what games you are playing, for how long, with whom and what you’re spending in those games. And they are updating their best guess of your lifetime value every single day. They’re looking at one billion customers a day and coming up with better estimates, then using that information in all kinds of surprising ways—not only should we sent this targeted email or not, but even more strategic things like when they launch a new game. Instead of rewarding the supplychain people on the basis of how many units they ship—that’s old school— let’s do it on the basis of how much CLV they elevate. So, let’s look at the new customers who are acquired through that game and see how valuable they ended up being. Or, let’s look at existing customers and see how much the existence and their purchase of this new game makes their CLV even higher. Let’s judge the products we offer, the services that surround them—pretty much everything we do—on the basis of the value of the customer.

Example number two would be Starbucks. Again, [they are] a company that came from very product-centric roots that was all about the right roast. It was all about the right ambiance within the store. But they’ve gotten so smart and they are leading the way when it comes to digital and mobile, because they recognize that some customers are more valuable than others. And [they] use that information to really optimize the overall experience, to come up with the parameters of the loyalty program and so on, with this idea that not all customers are created equal. Those are just two examples. I wish I could say most companies were like them, but they are the exception. But, more and more, slowly but surely, they’re becoming the rule.

Snow: This might be oversimplifying things, but in general when you apply some of these more rigorous customer-valuation techniques, do you find that the companies have been undervaluing their customer base? Or do they sometimes expect too much for them when the hard analysis is applied?

Fader: Obviously, the answer is that it depends. In the examples I’ve done so far, I’ve found that in a contractual setting—whether there’s a formal subscription or there’s an account and the company knows when the customer is leaving—it’s quite accurate. I have a new paper where we took the publically available information for Dish Network and for Sirius XM Satellite Radio. We did this bottom-up valuation and came up with numbers that were very, very close to what the market was saying. And some of the newer work that I’m doing now in the non-contractual setting—if we’re talking about a retailer or some kind of travel service or a zillion different companies out there, most companies that don’t have formal contracts—then, we find that companies tend to be undervaluing those customer assets or, therefore, the overall value of the company.

Again, it’s early and I don’t want to say this as a global statement, but in the non-contractual setting—where people are just making transactions over time and the rates of transactions vary and we don’t know when people are dropping out and the size of the transaction is very—there’s a lot more noise than there is in the contractual setting, when we know when people are going to renew and the amounts are often the same.

Snow: Applying a private equity screen to this, a bidder of a company armed with better information and a better predictive model for the customers is going to have an edge in a competitive bidding situation.

Fader: There’s no question about it. And there’s been a number of PE firms that have been watching my work in a way that’s—one might say creepy, but I’d say, really, it’s wonderful. It’s terrific. Again, I’m just a lowly marketing professor and to see people in private equity seriously caring about what I’m doing, taking the models and implementing them, maybe tweaking them in different ways, applying them in different ways, is very gratifying.

But, in the non-contractual setting, what are the right kinds of methods? The metrics, rather. There are still a lot of unanswered questions, not to mention the practices of people actually trying to use this stuff and where it fits within the organization and so on, where we’re barely started on it. But it’s really exciting to see where all this might go.

Snow: We’ll close with a bit of a crystal-ball question. Look 10 years into the future in the retail and consumer landscape.

Fader: The big problem with retailers today is that, when a particular customer walks in the store or, even worse yet, walks to the counter to pay for something, they have no idea who that person is. They have no idea of the value of that customer and, therefore, what they should be recommending or how they should be treating them differently. That is changing.

What I see in my crystal ball is that companies are going to have a much better recognition of the value of and the differences across their customers, not only when they walk in the store or when they’re checking out, but even during these offline interactions—when customers are posting reviews on social media or complaining about something on Twitter. Companies are going to know that this is a good customer [and they] better jump on it. So, the customer here is so-so. If we can get the bandwidth, sure, we’ll deal with them, too. But they can wait. To have that ability to be able to sort customers out—that’s happening.

Register now to watch this video and access all content.

It's FREE!

  • I agree to the Privcap terms of use and privacy policy
  • Already a subscriber? Sign In

  • This field is for validation purposes and should be left unchanged.