Propensity score methods have found wide application in a variety of areas, including medical research, economics, epidemiology, and education, where randomized experiments are difficult to perform or raise questions of ethics, and thus answers often must be obtained from observational (nonrandomized) data. Thus far, there have been few applications to marketing situations, even though the effects of many interventions (e.g., advertising, promotions) are "assessed" using generally inappropriate techniques such a regression, data mining, etc. This talk will present this new approach, and show how the more standard predictive methods can be used appropriately as a supplement. A real success story will also be presented.