A stock assessment requires a scientist to gather biological and fishery data in order to model the dynamics of a wild fish population. From this model they can estimate the stock biomass and subsequently provide advice to fisheries managers on future recommended catch or effort. Fisheries vary in complexity and data richness, and often model assumptions have to be made beyond those that can be informed by traditional datasets. This talk will explore three case studies in which expert elicitation—a scientific consensus methodology—has been used to inform model assumptions where data were lacking. This involved lead scientists seeking advice from those with experience "on the ground" (or on the water) in the fishery to better understand its dynamics. The stock assessments of east coast saddletail snapper and crimson snapper incorporated interviews with key fishers to better understand targeting behaviour in a multispecies fishery. The same stock assessments sought expert feedback using a custom, interactive web tool (built in RShiny) to characterise the selectivity of undersized fish. This profile was used to generate length distributions that represented the entire population, not just those for which data were collected. Now, the current assessments on three prawn species are being aided by an industry-wide survey on the spatial distribution of historical catch, used to fill in blind spots in the data that were generated by closures in the fishery. These three case studies are examples of methodologies used to complete narratives in fishery dynamics and histories where conventional data sources are absent.