We analyze the surplus extraction problem in a mechanism design setting with consideration sets. We study a bounded rationality version of a general mechanism design environment with correlation in which the agent evaluates only a subset of types as possible deviations. We call these subsets the agent's consideration sets. We identify the inverse consideration sets as the key elements that determine whether full extraction is feasible in this setting and characterize the conditions beliefs need to satisfy to guarantee full surplus extraction. These conditions require the beliefs of each type to be separated from the beliefs of types in his inverse consideration set only. This relaxes the independence condition in Crémer and McLean (1988), which remains sufficient in our setting. Finally, we discuss some applications and limitations of our model.

We study a monopolist’s product line design problem with search frictions. Consumers only evaluate a random subset of products in the menu, limiting the monopolist’s ability to perfectly match contracts to consumer types. This creates a tradeoff faced when expanding the product line between extracting more rents from different consumer types and increased matching costs. We show that when consumers are limited to seeing a single random product out of the menu, then the optimal menu for the monopolist always contains a single offer. When consumers observe more than one product, we show that a balanced menu where all products are seen by a consumer with the same probability is never optimal. The monopolist rather has an incentive to “bias” the menu so that some of the products are observed more often than others. Using an unbalanced menu has an impact on the quality provided to low valuation consumers, either reinforcing or reducing the distortions generated by asymmetric information. We discuss the consequences on quality provision, as well as the welfare effects of these distortions.