I had to go back to my notes in order to remember all of Hevner’s seven characteristics of good design science research, so that wasn’t so good. Maybe I can remember it better if I tell it as a story of the research process: We want to make an artifact that provides utility both to the business and to the research community (this is the research contribution). We evaluate our solution in order to determine utility and applicability, and we provide a research contribution by solving a heretofore unsolved yet important business problem (or by providing a better solution to a solved problem) in a scientifically rigorous (but not too rigorous) manner. And we want to communicate our results to both practitioners and scientists. Don’t try to remember the numbers from the paper, remember the story I told above, keeping in mind that design science occurs in a real business environment and yet is science, too, and so we have two sets of criteria and two audiences. That and that design science is concerned with utility, and not understanding.

I had trouble remembering how you evaluate non-instantiation artifacts. It goes like this: you evaluate them by how well they help you to design instantiations, essentially. I would guess that this means: evaluate constructs by how well they allow you to model; evaluate models by how well they allow you to build methods and instantiations; and methods by how well they help you design appropriate instantiations. The evaluations go from the general to the specific (assuming that constructs are the most general and instantiations are the most specific).
I should add that the symbiotic relationship between behavioral science and design science also goes behavioral -> design: behavioral science develops the theories and expands the IS knowledge base that design science uses.
I should talk about how design science has two audiences: practitioners and scientists. And in this way it is a way (like action science) of addressing the rigor-relevance dilemma.

I feel like this was a little short. I probably needed to expand the evaluations bit and the rigor bit. I could also expand the communication bit.