univ of maryland

Forecasting the Sales of Music Albums: A Functional Data Analysis of Demand and Supply Side P2P Data

Il-Horn Hann
Associate Professor of Information Systems
RH Smith School of Business
University of Maryland

November 20, 2009

Alter Hall 405, 1000am – 1130am

Abstract

We predict the sales of music albums by utilizing demand and supply side P2P data using a functional data
analysis (FDA) approach. We find that the characteristics of the functional form of downloading behavior explain
first-week sales by more than 60% after controlling for album characteristics. By updating our forecasts from 4
weeks to 1 week prior to the album release date, we examine the dynamic changes across different quantiles of
the sales-distribution for the demand- and supply-side P2P data. We find that the gap between downloading
effect on sales among high-quantile vs. low-quantile albums reach the highest level one week before the release
date.

For a copy of the full paper, click here.

Measuring the Value of Social Dynamics in Online Product Ratings Forums

Wendy Moe

Associate Professor of Marketing
R H Smith School of Business
University of Maryland

October 16, 2009

Alter Hall 746, 1130am – 100pm

Abstract

Extant research has shown that consumer online product ratings can significantly influence
product sales. However, these ratings have also been shown to be subject to a number of social
influences. In other words, posted product ratings not only reflect the customers’ experience
with the product but also reflect the influence of others’ ratings. The objective of this paper is to
model posted product ratings in an effort to measure the impact of the social dynamics that may
occur in a ratings environment on both subsequent rating behavior as well as product sales.

Our modeling efforts are two fold. First, we model the arrival of product ratings and separate the
effect of social influences from the underlying or baseline ratings behavior. Second, we model
product sales as a function of posted product ratings. However, rather than simply modeling the
effects of observed ratings, we decompose ratings into a baseline rating and the contribution of
social influence. From this model, we can measure the overall sales impact resulting from
observed social dynamics.

We show that ratings behavior is significantly affected by previously posted ratings. We further
show that the effect on sales resulting from this social dynamic is significant but relatively small
compared to the effect that ratings have when they represent an unbiased and independent
evaluation of the product. With the increased popularity of online discussion and ratings forums,
many marketers have been investing in efforts to moderate these conversations or to contribute
comments of their own, effectively biasing the sentiments expressed in the online forum. Our
results show that while these efforts can affect sales, their effects are limited and short-lived.