Online retailers are increasingly providing service technologies, such as online recommendation agents and live help, to assist customers with their online shopping. Despite the prevalence of these service technologies and the scholarly recognition of their importance, surprisingly little empirical research has examined the fundamental differences among service technologies. Consequently, little is known about the contingency factors that may favor the use of one type of service over another. In this paper, we propose the Model of Online Service Technologies (MOST) to theorize that the capacity of a service provider to accommodate the variability of customer inputs into the service process is the key difference among various types of service technologies. We also theorize the contingent role of task complexity with respect to changing the effectiveness of service technologies. We then empirically investigate the impact of service technologies that possess different capacities to accommodate customer input variability on the outcomes of efficiency and personalization, the two competing goals of service adoption. We posit two types of input variability in the service process: Service Provider-Elicited Variability (SPEV), where variability is determined in advance by the service provider; and User-Initiated Variability (SPEUIV), where customers determine variability in the service process. In our experimental study, we find that SPEV technologies are more efficient, but less personalized, than SPEUIV technologies. However, when task complexity is high, the superior effect of SPEV technologies over SPEUIV technologies in efficiency becomes less prominent, whereas both SPEV and SPEUIV technologies are perceived to have higher personalization. The results of this study further our understanding of the differences in efficiency and personalization among various types of online service technologies. They also serve to inform practitioners as to when and how to implement these technologies in the online shopping environment for the sake of improving efficiency and personalization for customers.