Publish-Discover-Compose paradigm of Service-Oriented Computing (SOC) presents a challenge on service applicability. Services are not just for predefined clients, rather for potentially many unknown clients. Hence, published services should be highly adaptable to various service clients and contexts. For that, service variability must carefully be modeled by considering the unique computing paradigm and requirements of SOC such as dynamic discovery and composition of services. Current SOC approaches to modeling services largely focus on defining business processes and service components without considering service variability in sufficient details. In this paper, we first compare the variability on conventional applications and the variability on SOC. Then, we identify four types of variability on services. For the types of service variability, we present a method to model service variability and design adaptable services. Using our proposed framework, we believe the applicability and reusability of such services can be greatly increased.