Leveraging Online Consumer Interest Tracking Data in Market Response Modeling

This dissertation develops techniques that use the information from online tracking data for analyzing market response. In theory, the observed market response originates from latent characteristics of the market such as consumers’ preference for products and features and the competitive landscape. Understanding these latent characteristics is essential in making high quality marketing decisions. However, finding a reliable and inexpensive proxy for them is a challenge. We explore the possibility of using insights from “big data” sources to better identify these latent characteristics. We apply our techniques to analyze the market for automobiles in the US. In Essay 1, we explore the potential of using trends in online searches for feature-related keywords as proxies for trends in the relative importance consumers place on the corresponding features. The relative importance consumers place on features may vary over time due to factors beyond the control of marketers (e.g., shifts in economic conditions, advances in technology). We make the baseline attractiveness of 70 top-selling automobiles a function of Google Trends indexes for five common features: fuel efficiency, acceleration, body type, cost to buy, and cost to operate. We find strong empirical evidence supporting the notion that the evolution of feature search intensity contains genuine information about shifting consumers’ tastes. In essay 2, we propose a model that identifies (1) the position of products on a latent perceptual map, (2) the consumer segments, and (3) the ideal point of the preference for each segment. The product positions are inferred using a novel approach using big data on online consumers’ activities. We show that our proposed approach performs better than alternative approaches in identifying latent product positions.