Developing adaptable systems is still a big challenge in software engineering. Different reference architectures and systematic approaches have been proposed to address this challenge. Several of these approaches are based on goal models, given their suitability to express and reason on alternative behaviors. In this paper we intend to provide a basis for comparing architectures described in goal-based models in regard to their adaptability. This way, different approaches to improve adaptability may be compared based on the resulting architectures. To do so we mapped two adaptability metrics onto i* models and developed guidelines to define the adaptability of individual elements, based on the extra information provided by i* models. We applied these metrics in a healthcare system to illustrate the comparison of architectures.

Peer Reviewed