Information systems become ubiquitous. This opens a large spectrum of possibilities for end-users, but the design complexity is increasing. So domain specific languages are proposed sometimes supported by appropriate processes. These proposals are interesting but they are under-validated. Even if validation is a difficult task, which requires specific knowledge, we argue that validation should be systematic. But many problems remain to be considered to achieve this goal: 1) computer scientists are often not trained to evaluation; 2) the domain of information systems design validation and evaluation is still under construction. To cope with the first problem, we propose to capitalize evaluation practices into patterns so that they can be reusable for non-specialists of validation practices. For the second issue, we propose an environment where evaluation specialists and engineering methods specialists can work together to define their common and reusable patterns.