This article reports on our experience of collecting language data from informants in video-conferencing settings under a research design originally developed with face-to-face interactions in mind. We had set out to investigate whether individual stylistic features persist in different modes of textual production and designed a complex set of data-collection procedures, which we then adopted for use in a fully virtual environment with 112 informants, who were asked to provide language samples in eight discourse types. We conclude that the unintended shift to virtual settings had only a minimal impact on the volume and quality of the data. While the process was occasionally afflicted by IT-related technical issues and made extra demands on the data collector, it also created opportunities, notably around the management of interactional power asymmetries. Additional benefits included instant troubleshooting during data handover sessions and the ability to recruit participants who would not have been able to travel to face-to-face sessions.