Despite the recent growth of interest in the interactional construction of research interviews and advances made in our understanding of the nature of such encounters, relatively little attention has been paid to the implications of this for interviewer training, with the result that advice on interviewing techniques tends to be very general. Drawing on analyses of a feature of research interviews that is usually treated as analytically insignificant, this article makes a case for more interactionally sensitive approaches to interviewer training. It focuses on interviewer recipiency in a database of over 40 research interviews conducted by academics and research students to show how apparently insignificant shifts in receipt tokens can have important implications in terms of the developing talk. The implications of this for researcher training are discussed and the article makes recommendations for ways in which attention can be drawn to the discoursal dimension in interviewing practice.