Businesspeople understand that not all customers are created equal—the 80-20 rule suggests that over time a small percentage of a company's customer base can generate a high percentage of its sales and profit. Models for calculating customer lifetime value are built on just such a premise.
But new research is starting to look at customers whose value is not as readily apparent, and where CLV calculations break down. In a recent working paper, Harvard Business School professor Sunil Gupta calls them "free" customers—think of buyers at an auction. Traditionally auction houses make most of their profit from fees paid by sellers; buyers don't pay fees. So although buyers are a necessary ingredient to the deal—no buyers, no sellers—their value is more difficult to quantify. To the auction house, is one buyer worth four sellers? Is one buyer worth one seller?
The answer is critical for the auction house, which must determine how to allocate marketing and other expenditures between buyers and sellers to attract new business.
As more job seekers sign on to Monster.com, more employers are willing to be paying customers for the firm.
Gupta's work provides a model for determining this value and is related to research being done by colleague Andrei Hagiu and others into the dynamics of multi-sided markets: platforms that serve two or more distinct groups of customers who value each other's participation. Just how that participation is valued is a question Gupta's research begins to answer.
Sarah Jane Gilbert: Why do traditional customer lifetime value formulas break down in a networked setting, and how does your model address those shortcomings?
Sunil Gupta: Traditional models of CLV estimate a customer's profit potential based on the purchase behavior of an individual customer. These models completely ignore the interaction among customers.
Consider a firm such as eBay that has two sets of customers—buyers and sellers. EBay generates almost all its profits from sellers through commissions and listing fees. Buyers do not provide any direct profit to the firm. However, without buyers, the firm would have no sellers and vice versa. This kind of situation, which is called a two-sided market, is common in many industries such as real estate and employment services.
A traditional model of CLV will not be able to estimate the worth of a buyer. In our paper we address exactly this question. We create a model which captures these "indirect network effects" where more buyers potentially attract more sellers and vice versa.
Q: How do "free" customers like the buyer noted earlier generate value for a firm?
A: A buyer on eBay does not provide direct revenues or profits to the firm but brings in more sellers and increasing numbers of transactions. As more job seekers sign on to Monster.com, more employers are willing to be paying customers for the firm. It is this indirect effect that generates value for a firm.
Q: Does customer value stay the same over time? Does it increase or decrease?
A: Customer value changes over time, and in our data we find that the value is increasing over time. In general, we expect customer value to initially increase as the firm grows and later decline when the firm reaches a critical mass or maturity.
Q: What are the practical applications of your research for companies in networked settings? In particular, what is the role of marketing in an environment with strong network effects?
A: Our research provides guidelines to firms such as eBay or Monster.com on how much they should spend to attract a new buyer or a job seeker. It also shows how a firm may want to allocate its marketing expenditure between buyers and sellers.
Q: Are there some networked environments where this model would work better than others?
A: So far we have focused on two-sided markets where a firm has two sets of customers who interact with each other. There are many other situations (e.g., MySpace) where the nature of customer interaction is much more complex. Our model will not be applicable in those settings.
Q: Was there anything in your findings of the data set you examined that surprised you?
A: We were surprised to find that, at the current time, in our data the value of an additional buyer is almost the same as the value of an additional seller even though the ratio of sellers to buyers is almost four to one.
Q: Will you continue with this research? What are you working on next?
A: I am very excited about this area. I am currently working on understanding and modeling complex network structures such as those of MySpace. Here the issue that we are grappling with is the tangible and intangible value of customers. In other words, customers provide tangible value to a firm through direct purchases but they also provide intangible value through network effects or word of mouth. It is quite possible that some customers have low tangible but high intangible value. Traditional models would label such customers as low value and would miss a huge opportunity for a firm.
