The recent emergence of online feedback and sampling mechanisms has created new avenues where consumers can learn about information goods such as books, music, and movies. An increasing number of studies evaluate the impact of these technological innovations on purchasing behavior. These studies often face estimation issues such as unobserved heterogeneity and simultaneity. This paper reviews the strategies commonly implemented to address various sources of endogeneity and proposes the use of a difference- and system-Generalized Method of Moments (GMM) estimator. The paper also stresses the need to consider the pattern of sales over time. As an empirical example, we explore the relationship between individual track sales and sampling of songs on the radio and MySpace. Our results suggest that while radio exposure continues to be an important predictor of song sales, online sampling of songs has a nearly equivalent effect on sales.