There is a tension at the heart of theory development. On one hand, new theory must acknowledge and build on existing theory so that it is relevant to current debates and can make a focused contribution. On the other hand, new theory must push boundaries beyond existing understandings. And when it develops new perspectives, it could potentially make existing theory less relevant. This tension is reflected in the coding process when analyzing qualitative data. It manifests as a dilemma, in particular: To what degree should the coding process, and subsequent category-building and theorizing be guided by existing theory? Versus, to what degree should coding and theory built on it be allowed to emerge via deep appreciation and close adherence to the first order perspectives of participants? This case study examines this dilemma in the context of a supervision dialogue between a doctoral candidate (Fernandes) and his doctoral advisor (Heracleous) that took place in February 2016, and centered on a coding document that the candidate had produced. The case highlights that one potential approach is to steer a middle path. That is, to find a way to acknowledge both existing theory and participants’ first order perspectives. This can be accomplished via being conscious of, and respecting, the “double hermeneutic” of social science, a concept advanced by sociologist Anthony Giddens. Furthermore, research students are advised to reflect on this dilemma, clarify how they have addressed it, and be ready to defend their methodological choices.