Extracting dimensions of quality from online user-generated content

摘要

The quality of a product may be an important driver of consumer satisfaction, competition, and long-term brand success. Identifying the right dimensions of product quality may be central to devising segmentation and positioning strategies, rankings brands, creating advertising content, improving current products, or designing new products. User-generated content, such as product reviews, may identify quality. Data in product reviews is analyzed across fifteen firms in five markets over four years to extract the dimensions of experienced quality using Latent Dirichlet Allocation. The face, external, and predictive validity of these dimensions is explored. Results suggest that a few dimensions may be enough to capture experienced quality, have good correspondence to other metrics of quality, and serve as reasonably good predictors of earnings and stock market returns. Dynamic analysis may enable tracking the importance of dimensions and of competitive brand positions on those dimensions over time.