Software architects are often confronted with a lack of crucial data and knowledge at the beginning of the development of new systems. As a result, architectural decisions in the early phase are often based on assumptions and guesswork. At the same time, the resulting architectures must not only meet current requirements, but also be able to deal with future uncertainties.

The presentation shows how data-driven decisions can serve as a compass for navigating through the uncertainties of software design and how this can free you from mere assumptions. In the following, concrete methods and procedures are presented that make it possible to make well-founded decisions despite missing or incomplete data. In addition, the presentation will show how architectures can be developed that are robust, scalable and future-proof in their design, even if not all the necessary data is available.

Practical examples will be used to show how the analysis of real user data and patterns leads to better architecture decisions in the long term. The lecture also presents strategies for proactively dealing with uncertainties in software design through the use of data and flexible modeling and adaptation techniques. In this way, forward-looking architectures can be designed even in data-poor scenarios.