Volcanic eruptions are multi-hazard events that cause a variety of impacts to exposed societies. One challenge scientists face is communicating intricate interactions between various eruption hazards and societal elements. Complex eruption scenarios that include multiple likely eruption hazards have been found to be an effective tool for communicating this information. These eruption scenarios are of relevance to end-users because they provide a more in-depth picture of a potential future eruption than those analyses that focus on a single eruption hazard. However, deep uncertainties within the analysis must be addressed to unravel the multifaceted interactions between an eruption and societal elements for the development of scenarios. Deep uncertainty manifests when probabilities of an event are poorly constrained, unknown, or unknowable. How can volcanic risk analysts provide appropriate decision support when fundamental probabilistic relationships are unknown or at best poorly understood? What if you must rely on limited information or case studies to develop scenarios? What are the implications if fundamental assumptions or scientific data the analysis heavily relies on turn out to be incorrect? These are some of the challenges volcanic hazard / impact / risk analysts must overcome to develop multi-hazard volcanic eruption scenarios and impact assessments. Here, we propose a transparent and structured framework for scenario development that provides guidance for analysts to not only develop robust hazard, impact and risk scenarios in the face of deep uncertainty, but to also ensure that they accurately report and communicate findings to end-users and the research community. Our approach borrows concepts from traditional gap-analysis methodologies where we identify and rank the degree of scientific knowledge the scenarios are based on. We demonstrate how this process has been used to develop seven new eruption scenarios in the Auckland Volcanic Field.