Exploring the impacts of a recommendation system on an e-platform based on consumers' online behavioral data

This paper investigates the impact of a recommendation system (RS) on various consumers of a kitchen-sharing platform as regards process efficiency and consumer satisfaction. RS effectiveness is determined on the basis of adequate observation of the historical behavior of consumers, which is based on their activity on the platform. To formulate our hypotheses, we considered the characteristics of the platform. We used propensity score matching and difference-in-differences methods to test our hypotheses, and the novel causal forest approach was used to ascertain the robustness of the results. Our findings suggest that consumers who adopted the RS experienced a 15 % increase in session count and a 2 % increase in purchase intensity. However, their processing efficiency decreased by 29 %. This may be because the RS of our emerging platform induces consumer behavior to spill over to additional stores and products, prolonging the decision-making process. We also discovered that the RS has limited influence on consumers who have developed ingrained routines with a high level of platform stickiness. For consumers with high store diversity, the RS not only induces their behavior to spill over into other stores but also generates an increase in consumption desires and orders, although these effects are temporary.