Background: Faced with the challenge of how research can help medical care cross the quality chasm separating what we think we know and what we actually do, new approaches and designs for research are needed. We especially need research that helps us understand problems in implementation and how to improve care. Since care and financing changes occur at an increasingly rapid pace, they provide huge opportunities for learning if we can only build and fund appropriate parallel research studies. This paper will describe such a study of a major statewide initiative to change the financing and care for depression in Minnesota. Methods: The DIAMOND (Depression Improvement Across Minnesota Offering a New Direction) study proposal was developed in conjunction with a similarly named collaborative plan for changing the way all health plans and the state paid for primary care of depression and for facilitating care changes that would implement evidence-based care. The implementation was planned in a way that would be most feasible for stakeholders while also being compatible with a sophisticated quasi-experimental research design that involved staggered implementation with multiple baseline measures of practice systems, care processes, and patient outcomes. It also featured evaluation of changes in worker productivity and in healthcare utilization/costs while also measuring the extent of penetration through the RE-AIM framework (Reach, Effectiveness, Adoption, Implementation, and Maintenance). Results: The research proposal to NIMH measurements before the changes began. Study measures required extensive cooperative efforts by the payers and medical groups, which were made possible by the ongoing integration of the initiative and the research and because all participants recognized the value of the measurements for assessing the effects and aiding decisions to maintain and extend the payment changes. Conclusions: It is possible to fund and evaluate even extensive change projects if developed collaboratively and flexibly with operational leaders in such a way as to provide information they value. Luck helps too.