An Analysis of Real World TV Advertising Tests: A 15-Year Update

Ye Hu
University of Houston

Leonard M. Lodish
The Wharton School of the University of Pennsylvania

Abba M. Krieger
The Wharton School of the University of Pennsylvania


It has been more than 10 years since Lodish et al. (1995a) published their meta-analytical study of IRI (Information Resources Inc.)
BehaviorScan
®'
s
(BSCAN) split cable in-market TV advertising experiments. This article partially updates that analysis and provides further managerial implications of the new results. [This study is a "partial" update of Lodish et al. (1995a) because of limited resources. In particular, the study by Lodish et al. (1995a) included detailed survey information about the advertising, marketing, competitive strategy, and tactics for each of the advertising campaigns, and extensive product competitive situations during each of the BSCAN tests. Such information cost over $1 million to gather. These resources were not available for this analysis.] From the original analysis, Lodish and colleagues reached a number of conclusions: some reinforced common beliefs about TV advertising-such as new-product TV advertising has greater impact on sales than TV advertising for established products; others were surprising and counter to conventional wisdom in the advertising industry-that increasing advertising budgets does not increase sales in general, and that there is not a monotonic relationship between increasing TV advertising and increasing sales. The in-market experiments used in Lodish et al. (1995a)'s study were completed between 1982 and 1988. Changes (e.g., audience segmentation; a typical TV advertisement used to last 1 minute, now it is usually 30 seconds in length; increased media fragmentation) during the past decade and a half might affect both the effectiveness and the managerial implication of weight increases of TV advertising.