The design of systems with better performance is a real need to fulfill user demands and generate profitable Web applications. Understand user behavior and workload they produce on the server is fundamental to evaluate the performance of systems and their improvements. User reactivity, that is, how the users react to variable server response time, is usually neglected during performance evaluation. This work addresses the use of reactivity to improve QoS of Internet services. We propose and evaluate new admission control policies. We designed and implemented the USAR-QoS simulator that allows the evaluation of the new QoS strategies considering the dynamic interaction between client and server sides in Internet services. The simulation uses a TPC-W-based workload and shows the benefits of the reactive policies, which can result in better QoS. The experiments show the proposed reactive admission control policies lead to better response time rates, with a reduction from 15 to 50%, preserving the user satisfaction metric.