The Design Options for the Search for Employment (DOSE) project, funded by the U.S. Department of Health and Human Services, involves creative, new thinking about how best we might evaluate not only the effectiveness of job search interventions overall but more importantly the relative effectiveness of specific components of job search for specific groups of particular policy interest.  In brief, the project is charged with creating an evaluation design that can answer the “what works best for whom?” question for job search strategies assisting low-income Americans.  To that end, in partnership with staff at the Office of Policy Research & Evaluation, a team at Abt Associates is crafting evaluation options that consider individual- and site-level randomization as well as spillover effects that accrue to society, outside of the purview of conventional evaluation research.  The paper proposed for presentation at APPAM will present results of this innovative design work, discussing both design and analytic strategies for delving within the “black box” of “what works?” with impact design tools, rather than relegating the opening of the black box to implementation researchers or ex post maximizing of available data.