Efficiently caching data at a client device

A merchant system computes various probabilities that visitors to a Web site will request individual Web pages of the Web site. The computed probabilities are then utilized to cache Web pages having the highest probabilities of being requested to a client device. The probability data may include aggregate probability data that defines the probability that any visitor to the Web site will request a Web page, customer segment probability data that defines the probability that customers in a particular customer segment will visit the Web pages, and/or customer-specific probability data that defines the probability that a specific customer of the Web site will visit the Web pages. Only Web pages having a computed probability greater than a caching threshold may be cached at the client device. Additionally, the Web pages may also be cached at the client device based upon the actual interaction with the Web site by a visitor.