In the growing field of information services, as in other industries, companies seek to maximize profits by matching customers with products based on the customers’ preferences and willingness to pay. But for an information services company, such as an Internet service provider, the demand curve is only one piece of an intricate puzzle. Economic models that analyze consumer behavior do not take into account the dynamics of the technology system delivering the service. By the same token, technology models that aim to optimize system performance pay little heed to the company’s bottom line.
So how should an information services company design its products — and the system that delivers those products — in order to maximize profits? Professors Costis Maglaras and Assaf Zeevi have studied such questions with research that combines a detailed model of consumer behavior with a simple model of system dynamics.
“If you look into the companies that deliver these services,” says Maglaras, “you’ll realize that the engineering function and the marketing and strategy functions are separate, and to a large extent they don’t talk to each other. Our work is trying to merge these two elements and propose models that are sufficiently rich to generate useful insights, yet not overly complicated. Should we be modeling the fine structure of the Internet? No, because then you just get bogged down in the details and you can’t say much about the economics of the system.”
Consumer behavior models typically make two basic assumptions: first, consumers vary in their preferences and willingness to pay, and second, faced with an array of products, consumers will make rational choices. When you add a complex service delivery system into the picture, you also have to consider such factors as congestion, bandwidth and service quality. Understanding customers’ sensitivity to those issues can help you determine whether it makes sense to offer a menu of differentiated services.
“If everybody has similar preferences in terms of delay,” Maglaras says, “even if they differ in terms of how much they’re willing to pay, the service provider should just offer one class of service. If on the other hand the market seems to be more heterogeneous — let’s say some people want to do voiceover IP using Skype and some want to do Web surfing and e-mail — then you’re better off offering different classes of service, because the people that want to use Skype have a different sensitivity to congestion versus the people that are just using the service to send e-mail to their friends.”
Once you’ve made the decision to segment your customer base, your delivery system must support that decision.
First, the system must have sufficient capacity to provide high-quality service to the higher-paying customers. Maglaras and Zeevi have developed a model that can help you determine the appropriate level of technology investment based on aggregate demand data. “If you do reasonable market research, you can estimate demand elasticities in certain industries — and there are studies that do that,” says Zeevi. “And since our model takes the elasticity as an input and spits out an insight like ‘Yes, the system will be very congested’ or ‘No, it won’t be,’ a service provider can plan accordingly.”
Second, in order to encourage customers to choose the higher-priced service, you must ensure that high-end customers suffer shorter delays than low-end customers—although the infrastructure supporting both types of service may be largely the same. Some companies deliberately provide slower service to low-end customers even though they could provide faster service at no extra cost. Recent work by Maglaras and Zeevi articulates in what situations this strategy would significantly increase profits.
“How do you segment the market in a way that people who are willing to pay more will indeed pay more?” asks Zeevi. “One possibility is to offer two classes of service, where the lower-end customers experience high congestion because resources are mostly used to serve high-paying customers.” In some cases this approach may not achieve sufficient differentiation, and high-paying customers may choose the lower-end service. “To prevent this from happening,” Zeevi says, “the service provider may artificially degrade the quality of the lower-end service — for example, imagine holding completed orders at a warehouse before shipping them out, or slowing down the rate of an Internet connection.”
Whether you opt for a tiered pricing structure or a one-rate-fits-all strategy, Maglaras and Zeevi’s research offers insights to help you strike the right balance between price and service quality. “If you’re offering good service quality at a high price,” says Zeevi, “you need to have ample capacity to actually make sure that service quality is being delivered.”
Costis Maglaras is the Philip H. Geier, Jr., Associate Professor of Business and Assaf Zeevi is the Gantcher Associate Professor of Business at Columbia Business School.
Costis Maglaras is a Professor at the Graduate School of Business at Columbia University in the division of Decision, Risk & Operations. His research focuses on quantitative pricing and revenue management, the economics, design, and operations of service systems, and financial engineering.
Costis received his BS in Electrical Engineering from Imperial College, London, in 1990, and his MS and PhD in Electrical Engineering...
Assaf Zeevi is the Vice Dean for Research and Henry Kravis Professor of Business at the Graduate School of Business, Columbia University. His research is broadly focused on the formulation and analysis of mathematical models of complex systems. He is particularly interested in the areas of stochastic modeling and statistics, and their synergistic application to problems arising in service operations, revenue management, and financial...
Read the Research
Costis Maglaras, Assaf Zeevi
"Pricing and Design of Differential Services: Approximate Analysis and Structural Insights"
Costis Maglaras, Assaf Zeevi
"Pricing and Capacity Sizing for Systems with Shared Resources: Scaling Relations and Approximate Solutions"