Determine the best attributes for an instant camera, narrow customer preferences for crossover vehicles, and identify ideal laptop bags for MBA students.
All could be standard product development research projects. But what if you could accomplish them interactively, over the Web, with unprecedented accuracy, and at a fraction of the time needed for traditional market research?
Still in its infancy, the Virtual Customer Initiative (VCI) has done that. And with the release last winter of the source code behind the initiative, its potential for revolutionizing market research and product development looms great.
A product of collaborative research by MIT Sloan marketing professors and doctoral students, the VCI uses computational algorithms, multimedia tools, and Web interactivity to obtain customer feedback.
The interactivity and immediacy of the Web hold advantages over traditional market research methods, but the VCI team is also employing new computational methods to reduce dramatically the number of questions required to pinpoint customer preferences.
“It's a quantum change,” says John Hauser, Kirin Professor of Marketing.
A typical market research project today could take up to six weeks and cost between $20,000 and $200,000, he says. The VCI cuts that time substantially, and it yields results that are far more accurate than those obtained through traditional means.
Led by Hauser, the VCI research team includes Ely Dahan, Duncan Simester, and Drazen Prelic, all MIT Sloan marketing professors, and the team has sought computational input from professors Andrew Lo, Robert Freund, and Tomaso Poggio, and others.
The VCI employs six web-based methods for eliciting customer product preferences. The methods range from the use of conjoint analysis, a traditional market research tool, to an interactive game in which respondents trade product concepts as securities.
The methods are suited for different goals, and in tests each has been shown to improve market research capabilities.
Using traditional conjoint analysis, the team tackled projects for an instant camera, crossover vehicles, and laptop bags, and on-line demos of those projects illustrate the power of virtual market research.
Customers click through screens in which virtual realistic prototypes are equipped with an array of features. Then, customers reveal their preferences by sorting cards or choosing among concepts representing combinations of features.
With those projects, researchers found the Web interface proved a valuable tool to advance traditional research. Still, the team was concerned that traditional conjoint analysis required customers to decide among too many options.
With its so-called FastPace method, the team began employing groundbreaking interior point calculations as a way to determine customer preferences in as few questions as possible.
Duncan Simester explains that the method approximates the range of feasible customer preferences using an ellipsoid. When the computer asks a question, the answer eliminates a portion of the ellipsoid, and the computer can refine the next question.
“The solution is to choose a question that slices up the space as quickly as possible,” he says.
The VCI methods promise better market research in part because they are easier and more interesting for customers.
Most importantly, the methods yield a greater response rate and greater accuracy than traditional methods. Research on MBA students' laptop bag preferences illustrates those advantages.
The VCI team gave 350 MIT Sloan MBA students $100 to purchase a laptop bag. Viewing realistic prototypes and features, each student clicked through options and selected a personalized bag.
The bags were priced below $100, according to the features selected. As an incentive the students received cash for the difference between the purchase price and their $100 allotment.
“The response rate was over 90 percent,” says Simester.
Obtaining such feedback through traditional means would have been time-consuming and costly, and the results would have been less valid.
A marketable commodity for those who refine it, the VCI source code is now public, with the development last winter of a VCI website.
The website offers demos, PDFs of research papers, and downloadable source code for the VCI methods. It's a bountiful resource for university and private researchers.
Why release the code, when the VCI has such marketable potential?
Ely Dahan says the team has an eye on the greater good — to initiate a dialogue among researchers, to encourage others to develop further and improve the VCI methods, and to propel the concept closer to its revolutionary potential.
“Someday,” he says of the still nascent tool, “this may be as common as building a spreadsheet or creating a PowerPoint presentation.”
The complex algorithms at the core of the VCI have applications beyond Web-based market research and product development. In the near future, says Dahan, they could be used to help us all make better purchasing decisions.
Imagine a portable wireless communications device, he explains, which records our choices and learns our preferences. This “Intelligent Advice Module (IAM)” could also interview its owner using Virtual Customer methods.
When facing a choice, the owner could turn to the device. The device would run algorithms, taking into account all prior choices, then communicate with firms about their offerings, and yield a decision consistent with the owner's preferences. Ultimately, the owner could come to view the IAM device as a trusted friend.
“It would have enough information to give you answers to important questions,” says Dahan. “Which car to buy, which schools to consider, which place to visit. It could be very effective.”
[ back ]