Service orientation has become an attractive solution to many software development and integration problems. To fully adopt the service orientation concept, business organisations need to learn about a new methodology for service architecture and employ it to their existing software systems. Such an effort is likely to pay off later when the service-based systems can accommodate changes in a more flexible way. However, since the benefits are not seen right away, this could deter business organisations, especially smaller enterprises, from service adoption. This paper presents an experiment to determine agility to change of a service-based system in comparison with that of a non-service-based one to check if the service-based version is more agile to change. The two experimental systems are in the context of an E-Leave application of a small software company in Thailand. The model we use to measure agility extends an existing software agility metric by bringing code complexity into the model. The experimental results show that the servicebased system is more agile to change than its counterpart, while code complexity can help give a better view of change complexity that affects agility. We also note that agility of the service-based system can be improved if the services are well-designed.