MIT's Virtual Customer Initiative (VCI) is a multidisciplinary research project developing and testing new theory and methods to improve the speed, accuracy, and usability of customer input to the product development process. Our ultimate goal is to drive a customer focus deep within a corporation by providing simple, easy-to-use, and effective web-based tools to obtain and use customer input.
The VCI is actively engaged in developing and testing new methods to elicit customer input with web-based data. We develop and implement new methods with "proof of concept" applications. These applications demonstrate that the methods are feasible and powerful. We publish the details of the methods and, when feasible, provide open-source code so that others might apply these methods in an enterprise setting. We are fortunate that many firms have begun to apply these methods effectively.
Some of the methods that we have developed include:
FastPace Family of Conjoint Analyses
- FastPace: an adaptive method for metric data that has proven to be more accurate than the industry leading adaptive conjoint analysis
- FastPace CBC: the first method for choice-based data that adapts questions to each respondent
- FPCBCi: an improved adaptive CBC method that uses prior information and expands polyhedral methods based on a formal error theory
- Support vector machines: Our colleagues at INSEAD have develop new optimization methods for question design and partworth estimation.
Cards Family of Preference Measurement and Conjoint Analyses
- Gards: a new method to identify "must-have" features that identifies the heuristic processes that consumers actually use. This method can identify a variety of lexicographic processing strategies.
- Cards: a stream-lined conjoint analysis method for full-profile ranks that shows respondents only those profiles that are consistent with an additive model. Cards reduces the measurement burden on consumers dramatically.
- Both Gards and Cards utilize a measurement task that is perceived by respondents to be simpler, more-natural, and more accurate. In addition, our colleagues at Columbia University have developed a related method based on metric data.
New Insights on Conjoint Measurement
- M-efficiency (managerial efficiency). These analyses suggest that for managerial decision making, traditional measures of efficiency are best replaced with a measure more directed at key decisions
- Utility balance. It is commonly believed that "hard" questions give more accurate data. We found this to be true for choice-based data, but for metric data, adaptive utility balance biases small partworths upward. We suggest methods to overcome these biases.
- Ideation game: We have developed a simple structure with incentives and self-monitoring that helps product development teams identify more ideas and better ideas. The method takes the form of a game in which respondents gain points by building on one another's ideas. This method has been applied commercially to great success.
- Information pump: In this game respondents are rewarded for thinking hard and providing information that is not redundant.
- Trusted advisors: We have developed Bayesian methods so that web-based advisors can select questions to ask respondents so that the advisors can gather information rapidly and provide better advice.
- Listening in: By monitoring trusted advisors (with respondent permission), we "listen in" to identify new product gaps that represent profitable new combinations of product features that are demanded by respondents but not currently available on the market.
- Information scoring: Information scoring is a "truth serum" that provides respondents incentives to tell the truth, even for subjective data.
- Incentive compatible games. Although still under development, we are experimenting with information games in which respondents monitor one another in a fun setting. These gains automatically provide the incentives for truth-telling.
Other Web-based Methods
- Stock-market-like trading systems to evaluate new product concepts
- "Configurators" that allow customers to select product features in attractive and easy-to-use formats
- Genetic algorithms for product design validation testing of successful commercial methods.
- Virtual concept testing using web-based methods.
- virtual customer methods published in the Journal of Product Innovation Management
- product development for the Handbook of Marketing
- innovation research for the Marketing Science Institute and Marketing Science
- conjoint analysis in honor of Paul E. Green
For more information on these developments (and other developments), examples of various applications, downloadable and server-based demos, open-source code, research papers, presentations, and links to firms implementing these methods, we welcome you to the Virtual Customer Web site.
|The Center for Innovation in Product Development, a sponsor of the Virtual Customer Initiative, unites industry representatives with leading research faculty to investigate the end-to-end product development process—from engineering concept to market launch and beyond.
Learn more >>