Commercialization of innovations frequently stumbles. A prominent recent example are the early (i.e., pre-3G) mobile phone-enabled Internet services, whose European take-up was slower than expected. To determine why, we build a structural model of demand for such services and estimate it using consumer-level panel data from a pricing experiment. The experiment allows a decomposition of the number of wireless connections into the number of needs—instances where a consumer would establish a connection if the price were zero—and the conditional probability of establishing a connection. We find that needs were plenty and potential consumer surplus several magnitudes higher than that attained. We find that pricing reduced usage substantially and explore potential reasons for the high prices.