Written records of spoken interaction are typically assumed to be adequate for the purpose they serve, often receiving minimal scrutiny from the institutions which consume them. In this article, we scrutinize the current practices of capturing spoken interaction in legal contexts in England and Wales, and highlight some of the often serious legal consequences that result. We ask five questions of record keeping in legal settings: (i) Is the record produced an accurate representation of the spoken interaction?; (ii) Do lay and professional participants have ownership? Answered by giving careful thought to the rights they may or may not have to their data; (iii) Who has agency, who’s ‘voice’ is represented in the recorded account?; (iv) Then, we ask how usable the record is; and (v) How resource efficient it is to produce and use. By asking these questions, we make visible the underlying assumptions about transcription adequacy—in doing so, we acknowledge and enable reflection on the process of capturing spoken interaction. We envisage this model to be applicable to a range of institutional settings.