Using Online Search Data to Forecast New Product Sales

This dissertation focuses on online search as a measure of consumer interest. Internet use is at an all-time high in the United States, and according to the Pew Internet & American Life Project, 91% of Internet users use search engines to find information. Consumers' choices of search terms are not well understood. However, we argue that people will focus their searches on terms that are of interest to them. As such, data on the search terms used can provide valuable measures and indicators of consumer interest in a market. This can be particularly valuable to managers in search of tools to gauge potential product interest in a new product launch. In this research, we develop a model of pre-launch search activity. We find search term usage to follow rather predictable patterns in the pre-launch and post-launch periods. As such, we extend our pre-launch search model to link pre-release search behavior to release-week sales - providing a very valuable forecasting tool. We illustrate this approach in the context of motion pictures. Our modeling framework links search activity to sales and incorporates product characteristics. Our results indicate consistent patterns of search over time and systematic relationships between search volume, sales, and product attributes. We extend our model by studying the role of advertising. This allows us to better understand the relationship between advertising and online search activity and also allows us to compare the forecasting performances of each of the two approaches. We find that search data offers significant forecasting power in opening-weekend box-office revenues. We further find that advertising, combined with search data, offers improved forecasting ability.