Software architectural analysis is critically important before a system is built or for system evolution. The article presents a methodology that the author has been developing and testing in evaluating the sensitivity of software architectures with respect to changes in key customer values and future requirements at Nortel's SEAL. The approach consists of a framework for modeling various types of relevant information and a set of architectural views for re-engineering, analyzing, and comparing software architectures. The article highlights the approach by applying it to the common problem of this panel. The information that one needs to add to the architecture for the analysis is described. The article also presents some potential results, lessons learned from empirical studies, and the current research works in progress.