The time-critical tasks on the Internet are increasing. The delays in these tasks can have severe implications on the Quality of Experience (QoE) of a service. Therefore, networks require smart user-centric resource management mechanisms to reduce the impact of these delays on QoE. For this, a better understanding of the user preferences with regards to service performance is a prerequisite. In this paper, we present user responses to the three different distributions of delays, occurring during shopping sessions on the Web. By keeping the overall waiting time of the sessions same, we show how the users respond differently to the different set of delays. We analyzed the user responses and found that, the users prefer small frequently occurring delays as compared to the long rarely occurring delays within a task-based session.