CAMBRIDGE, MASS., Oct. 4, 2017––Researchers, including MIT Sloan School of Management’s Alessandro Bonatti, have designed what they believe is an improved, quantitative way for big-data providers to price the information they sell to informed customers.
The pricing of information has long vexed many data miners and others who collect all sorts of information from a wide variety of sources about customers, from the credit histories of those seeking bank loans to the viewing preferences of cable viewers to shopping patterns of online consumers browsing through web sites.
The sometimes highly sophisticated and detailed data is known to be potentially lucrative, but sellers have often run into problems structuring their prices based on how much data individual customers already have – and how much extra information they may actually need and want.
“Data sellers, ideally, offer a menu of options they hope is attractive to potential customers,” said Bonatti, associate professor of applied economics at MIT Sloan. “What we’ve established is a quantitative, highly structured, revenue-maximizing menu of experiments that sellers must satisfy in order to achieve optimum sales of data products.”
MIT Sloan’s Bonatti, along with Dirk Bergemann, Douglass and Marion Campbell professor of economics and Computer Science, Yale University, and Alex Smolin, postdoctoral researcher in economics, University of Bonn, outline their findings in their new paper “The Design and Price of Information,” forthcoming in the American Economic Review.
In their new paper, the authors review their three main findings.
First, that sellers of data are constantly confronted with differences in buyers’ private beliefs in what information they need, introducing a novel, sometimes frustrating aspect to the selling of information. Second, that information is inherently rich and can be modified to suit the needs of individual customers. Third, that instrumental information and options are useful to buyers and, as long as the various options are carefully designed, they are profitable for the seller.
“We find that there’s often too much ‘noise,’ or bells and whistles, if you will, when data sellers offer alternative plans to buyers,” said Bonatti. “The inclusion of too much ‘noise’ is not a driver of profitability.”
The authors believe they have established a structured framework by which sellers can screen buyers and analyze the sale of supplemental information to various types of buyers, whether they’re “high-end” or “lower-end” buyers.
“As the collection and sale of data becomes big business, we provide some answers to how much information a data seller should provide and how he or she can better price access to the data,” said Bonatti.