Conservation of at-risk species often requires rapid status assessments and timely policy decisions. However, conservation policy implementation (e.g., species listings) may precede data availability and decisions are often based on sparse, uncertain, or potentially flawed information. Here, we present a case study on Humboldt martens (Martes caurina humboldtensis), rare and cryptic carnivores that occur in coastal forests of California and Oregon. Humboldt martens, and the analogous coastal distinct population segment of Pacific martens, were recently listed as state endangered (California, 2019) and federally threatened (2020), yet little is known about many aspects of Humboldt marten population ecology. We modeled population growth of a Humboldt marten population in northern California, incorporating empirically-derived demographic data under a variety of scenarios. All scenarios indicated substantial variability in population growth trajectories as an artifact of highly variable survival estimates. Our work exemplifies the issues of using sparse data to inform conservation decisions, and we provide a framework to improve outcomes of future research efforts of poorly-understood species. We suggest that increased collaboration, purposeful study designs, use of standardized protocols, application of emerging technologies, and data sharing via public repositories are reliable actions that can better produce biologically relevant results and inform future conservation efforts.