When developing complex software-intensive systems, it is nowadays common practice to base the solution partly on existing software components. Selecting which components to use becomes a critical decision in development, but it is currently not well supported through methods and tools. This paper discusses how a decision support system for this problem could benefit from a software ecosystem approach, where participants share knowledge across organizations both through reuse of analysis models, and through partially disclosed past decision cases. We show how the ecosystem architecture becomes fundamental to deal with efficient knowledge sharing, while respecting constraints on integrity of intellectual property. A concrete architecture proposal is outlined, which is a web-based distributed system-of-systems. Experiences of a proof-of-concept implementation are also described.