Evidence suggests that studies aiming to improve healthcare practice should be flexible and prioritise patient, family and clinician engagement. Video-reflexive ethnography (VRE), a form of qualitative research often employed in healthcare settings, is well-suited to these aims. VRE supplements ethnographic techniques with video-recordings of in situ practices, allowing practitioners to reflect on taken-for-granted practices. Its prioritisation of collaboration, affective entanglement, theory-driven analysis and flexibility – aligned with participatory and post-qualitative inquiry (PQI) – can facilitate reflexivity among researchers and participants for local practice improvement. Yet paradoxically, flexibility can hinder the predictability of impact, and demonstrating likely impact is crucial to securing research funding. This article offers practical advice to qualitative researchers facing this methodological challenge. Using three exemplars, we examine how differing onto-epistemological groundings, conceptualisations of participant engagement and researcher positionings affect the timing, predictability, scalability and transferability of each study’s impact. We show how prioritising affective engagement, flexible goals and collaboration can enable local healthcare practice improvement; prioritising theory generation via consultation can lead to traditional, more transferable, forms of impact. We share insights for researchers seeking to improve healthcare using methods inspired by PQI such as VRE. While predicting impact is fraught, optimising conditions for impactful VRE research can be accomplished by: foregrounding epistemology; prioritising affective engagement; aligning research and stakeholder goals; assessing timing and organisational readiness; and considering researcher and participant positioning.