It's obvious that automated testing is a long-term and costly investment. Accordingly, before the automation is started we need to answer the primary question - what tests are automated in the first place? And even more important question is - how to build primary acceptance test suite to validate the system under test?
SoftServe architectural group developed a methodology for creating an acceptance test suite for an application which is based on the quality attributes of the system (or the non-functional requirements). The methodology is based on the architecture definition technique, developed by the Carnegie Mellon Software Engineering Institute. I would like to share principles of this methodology, as well as real use cases.