Information Cascades in E-Commerce
As an increasing portion of the shopping done by consumers in the United States and around the world moves online, the concept of information cascades and what differentiates successful products from unsuccessful products becomes increasingly relevant. One of the major differences between shopping online vs in-store is the customer’s ability to physically see and feel the item. Without access to the physical product in front of them, consumers are, according to this study, relying most heavily on other customer reviews, picture reviews, and descriptions when making decisions about which products to buy (Fan et. al). In the paper, the authors were looking specifically on TaoBao, a Chinese e-commerce website and focused only on cleansers, a product they believe is difficult for consumers to get accurate reviews before buying and is a product that is used daily with experiences varying from individual to individual.
This pattern is relevant to what we learned in class about information cascades, which is observed when people tend to follow the crowd, or in this case, buy what others are buying or buy products that have a high number of ratings and also high ratings. The study found that negative reviews had a negative impact on the consumer’s decision making on whether or not to buy the product while high ratings encouraged consumers to buy the product. We can relate this to the general cascade model, where there were a total of 3 components with the first being the states of the product. In this case, either the product is good or bad (the customer is either satisfied or not with the product). the second component is the payoff, which takes on a similar meaning in this scenario. If the customer finds the product to be of satisfaction, they will receive a payoff. The third ingredient is signals, which can take the form of online reviews or perhaps word of mouth. As we saw in the text, Networks, Crowds, and Markets, “A high signal is more likely to occur if the option is good than if it is bad, so if an individual observes a high signal they raise their estimate of the probability that the option is good” (Easley and Kleinberg). This holds true in this model as well where if the product is truly good, an individual will find more positive reviews and raise the probability that they think the product will be of value to them.