Online user-generated content in various social media websites, such as consumer experiences, user feedback, and product reviews, has increasingly become the primary information source for both consumers and businesses. In this study, we aim to look beyond the quantitative summary and unidimensional interpretation of online user reviews to provide a more comprehensive view of online user-generated content. Moreover, we would like to extend the current literature to the more customer-driven service industries, particularly the hotel industry. We obtain a unique and extensive dataset of online user reviews for hotels across various review sites and over long time periods. We use the sentiment analysis technique to decompose user reviews into different dimensions to measure hotel service quality and performance based on the SERVPERF model. Those dimensions are then incorporated into econometrics models to examine their effect in shaping users’ overall evaluation and content-generating behavior. The results suggest that different dimensions of user reviews have significantly different effects in forming user evaluation and driving content generation. This paper demonstrates the importance of using textual data to measure consumers’ relative preferences for service quality and evaluate service performance.