Large-Scale Demand Estimation with Search Data

Tomomichi  Amano; Andrew  Rhodes; Stephan  Seiler

Working Paper | HBS Working Paper Series | 2019

Large-Scale Demand Estimation with Search Data

by Tomomichi Amano, Andrew Rhodes and Stephan Seiler

Abstract

In many online markets, traditional methods of demand estimation are difficult to implement because assortments are very large and individual products are sold infrequently. At the same time, data on consumer search (i.e., browsing) behavior are often available and are much more abundant than purchase data. We propose a demand model that caters to this type of setting. Our approach is computationally light and allows for flexible cross-price elasticities that are informed by search patterns. We apply the model to a data set containing search and purchase information from a retailer stocking almost 600 products, recover the elasticity matrix, and solve for optimal prices for the entire assortment.

Keywords: high-dimensional data; demand estimation; consideration sets; consumer search;

Language: English

Format: Print

34 pages

SSRN Read Now

Citation:

Amano, Tomomichi, Andrew Rhodes, and Stephan Seiler. "Large-Scale Demand Estimation with Search Data." Harvard Business School Working Paper, No. 19-022, September 2018. (Revised June 2019. Stanford University Research Paper, No. 18-36, 8-20 2018.)

In the summer of 2019, Yu Sasaki, Head of the Game Division of DeNA, a Japanese mobile gaming company, is evaluating various growth strategies for its recent game Othellonia. Sasaki needs to decide if he should focus on customer acquisition, retention, or monetization.

Launched as a private brand in 1980 to counter the increasingly brand-conscious consumer in Japan, MUJI offered beautifully designed, fairly priced, no-frills quality goods. The once modest private label brand with 40 products had expanded significantly by 2019 to more than 7,000 products with more than half its 975 stores outside its home market in Japan. It had even expanded into the service industry, opening hotels. President Matsuzaki of Ryohin Keikaku, MUJI’s operating company, was charged with reorganizing the product portfolio and prioritizing new initiatives, tasks complicated by the absence of a clear definition of “MUJI-ness,” the meaning of which had always been intentionally left open.

Faculty & Research

Large-Scale Demand Estimation with Search Data

Abstract

Citation:

About the Author

More from the Author