When researchers are interested in the experiences of couples, the mode of interview is typically considered a binary choice between separate individual interviews with each partner, or a joint interview with both partners together. That is, if interview mode is explicitly considered at all. In this article, we illustrate a reflective process undertaken to explore the role of interview mode in the production of knowledge. Our focus is the adoption of multi-level semi-structured interviews wherein couples were interviewed both jointly and individually in one visit. The paper is set out in two parts. In part one, the study context and how the mode of interview was conceptualized is considered, before describing the chosen multi-level interview design. In part two, how the mode of interview worked in practice is discussed. The triangulation of individual and dyadic level perspectives collected rich data. Despite the novelty of mode, the challenges encountered reflected familiar concerns with semi-structured interviews: characteristic match between interviewer and interviewee, recording tacit knowledge, moving beyond normative expression and balancing disclosure with interviewee well-being. The paper concludes with a consideration of our assumptions of what constitutes a “successful” interview and offers guiding reflective questions for researchers who are considering semi-structured interviews. Further research is needed to explore the impact of different interview modes.