"We propose to study the working of markets when consumers do not want to search all available choice options and when sellers recommend products based on what ""similar consumers"" have bought in the past. We argue that the current practice of on-line selling offers significant potential for selection mechanisms or limitations of what the consumer is aware of, prior to purchase. To study this issue, we propose and estimate a demand model that includes a formal model of consumer information search. This model accounts for limited product search and information effects on choice. Our proposed demand model can be estimated from a combination of browsing histories and ordinal demand data. Because these data are available for a wide cross-section of durable goods, the method we develop will likely have broader application than the strict confines of the current empirical context. The goal of developing this demand model is (i) to measure the severity of selectivity in self-motivated or seller-directed information search and (ii) to subsequently evaluate the consequences of selection for consumer welfare, and for market structure (e.g., concentration and firm competition). For instance, most product recommendation systems used in on-line selling create a feedback, between past sales and current sales, because popular products get recommended more often than unpopular ones. This polarizes product popularity beyond what it already was. Clearly such effects hold relevance when evaluating competition and entry, because in a feedback market where the big get bigger and the small get smaller, firms potentially offering new and innovative products find it difficult to enter. In sum, our study consists of (i) the development of demand models under imperfect product search, and (ii) the evaluation of the effects on consumer welfare and market structure of selective product search. We expect these aspects to be relevant in the study of (pan)-European markets."