among consumers and between consumers and brands. Social media users can now interact with brands directly through Facebook and Twitter. These new features make social media a very distinctive class of online WOM. In this paper, we formulate a random coefficient discrete choice model of consumer demand to study whether and how consumer engagement and attention on the Internet affect the consumption of carbonated soft drinks (CSDs). We model consumer attention and engagement on social media as goodwill in order to capture the carry-over effects of WOM’s impact on demand and combine two types of product level data: monthly CSD sales data and social media data. Our results suggest that the three types of social media messages all have significant and positive effects on CSD demand. In particular, indirect engagement on Twitter has the largest impact on consumer demand, followed by direct engagement on Twitter and consumer engagement on Facebook.