In this article we consider the 'impact case study' (ICS) as a specific kind of document, one which, as part of the U.K.'s Research Excellence Framework (REF), enforces a common template for the description and measurement of the social and economic effects of research in U.K. higher education. We track the development of an ICS describing anthropological research in tobacco control which, after many iterations, was not submitted as part of the REF. We ask 'what is impact?' in cases where anthropological research is based on principles of collaboration and serendipity rather than the mechanistic 'research > translation > impact > measurement' model which an ICS is expected to follow. What is included and what excluded by the strictures of such a model? We are generally supportive of the impact agenda, feeling that university resources and activities have a vital role to play in progressive social change. However, the way 'impact' is recorded, appraised and measured in an ICS only captures a small proportion of the effects of anthropological research, and encourages particular forms of public engagement while discounting others.