Consumers' purchase decisions depend on whether a product is perceived as a bargain or as overpriced. But how do consumers evaluate sales prices? The standard approach in economics, psychology, and marketing suggests that consumers' estimates are best described by a attribute-based or piecemeal strategy that integrates information about products in a linear additive fashion. Here, we outline and test an alternative theoretical approach from the categorization literature suggesting that consumers sometimes follow an exemplar-based strategy that relies on similarity to previously encountered products. We hypothesize that people switch between these two estimation strategies depending on the context they face. To test this hypothesis, we conducted an experiment in which 64 participants repeatedly estimated the market price of different consumer products (bottles of wine). In one condition, the product prices could be well approximated with an attribute-based strategy whereas in the other condition an exemplar-based strategy worked best. Results of a subsequent testing phase indicated that participants switched between strategies depending on the structure of the presented sets. These results show that people rely on different strategies to estimate market prices, which should influence people's consumption behavior. The results suggest that theories on categorization learning can provide a deeper insight into behavior in an economic context and allow predicting consumer behavior more accurately.