Challengers can successfully compete against dominant firms in network markets through innovation. However, the type of innovation a challenger should compete with depends on characteristics of the market and technology. I develop a typology of product innovation to model expected returns from different types of innovation. This model indicates radical innovation can be more profitable than incremental innovation, incompatible innovation more profitable than compatible innovation, and radical-incompatible innovation more profitable—and less risky—than incremental-compatible innovation. Sometimes more risk is more prudent, given the market in play.