The ROI of Electronic Media

From Questions to Answers
Media Mix Modeling and the Return on Advertising Investment

Kraig Schultz
Jon Swallen
Ogilvy and Mather, United States

Experience has shown that each of these three view-points, that of statistics, economic theory, and mathematics, is a necessary, but not by itself a sufficient, condition for a real understanding of the quantitative relations in modern economic life.  It is the unification of all three that is powerful. And it is this unification that constitutes econometrics.
Ragnar Frisch, 1933

The debate surrounding the use of econometric models to quantify advertisings impact on market results seems no closer to resolution today than it was ten years ago. Some declare that modeling has finally cracked the nut, and proves definitively that advertising has quantifiable short- and long-term impacts.1, 2 Models are Machine Dreams3 that lead us to the Promised Land4 of advertising accountability. Other authors declare that these machinations are mere hocus-pocus, and that the value of the insight they may offer is limited at best.5, 6 Yet still others simply throw up their hands in frustration at the complexities of the modeling process. Much of the confusion surrounding models, however, may be traced to a lack of clear objectives for many of the studies which employ this technique. With no explicit set of questions and hypotheses identified at the outset, it is nearly impossible to tell whether the assumptions underlying the model are tenable, whether the data was appropriate, or whether the technical aspects of the models estimation were correctly executed.