UCLA

This dissertation contains three essays that study the implications of online search activity for new-product marketing. Using the U.S. motion picture industry as a test case, the first essay examines the dynamic causal relationship between traditional media, consumers' media generation activity, media consumption activity, and market demand of movies. Consumers' media generation and consumption activities around movies are operationalized by the blog volume and search volume of those movies. I develop three separate models--a pre-launch period model, a post-launch period model, and an opening-week model--and examine the relationship between them separately for each period. As the focal variables are jointly determined, I introduce instruments and explain how to correct for endogeneity bias. I find that consumer searching activity is a key mediator between advertising, consumer blogging activity, and market demand.

The second essay examines the pre-launch advertising effectiveness of new products using online search indexes as the response variable. I model the relationship between the advertising schedule and the online search volume process during the pre-launch period of movies. The model incorporates consumers' willingness-to-search and the time-varying effectiveness of advertising as key elements that influence online search volume at specific times. The model is represented in a Bayesian dynamic linear model framework and applied to the U.S. movie industry. The empirical analysis reveals important features of consumers' pre-launch interest development for new products. First, consumers' pre-launch responses to advertising are substantially influenced by the timing of advertising. Second, advertising effectiveness varies over time as a function of past advertising outlay. Third, the time-to-launch effect and time-varying advertising effectiveness vary substantially across movies. The estimation results are used to suggest a more effective pre-launch advertising schedule.

The third essay provides an explanation of the varying predictive power of online search volume by examining the effects of consumers' quality perception on the search activity for and actual demand of new products. I hypothesize that both the perceived quality and quality uncertainty of a new product increase consumers' search activity for the new product, while only perceived quality positively influences the conversion of search activity into actual demand. Using the U.S. movie industry as a test case, I find that both perceived quality and quality uncertainty are positively associated with the pre-launch search volume of movies, whereas only perceived quality positively influences the conversion of pre-launch searches into opening-week revenue. Similar findings are maintained in the post-launch period analysis. These findings imply that systematic over-/under-prediction of market success may occur if managers use an online search index without considering the effects of quality and quality uncertainty of new products.