In a general on-line service environment, a service provider (SP) offers a set of service levels to customers at various prices. Higher priced service levels typically include a guarantee for faster response/completion times. The problem that an SP faces is pricing different service levels. Such a price schedule greatly impacts the rate of incoming requests, the mix of their service levels, and consequently the revenue of the SP. In this work we propose a data driven method for estimating the rate and the mix for any given price schedule. The data exclusively consist of the observed rate of request submissions and the actual choices that customers have made when presented with previously used price schedule(s). Given this characterization, the SP is able to provision resources and optimize scheduling for a given price schedule and ultimately choose the price schedule that optimizes the revenue. The problem is motivated by network service and utility computing environment applications.