Here, we combine a set of methodological approaches to define priority areas for proactive and reactive conservation management of poorly known species. We used the Brazilian endemic three-banded armadillo as a case study, a species once thought to be extinct in the wild and currently known by a few dozen records in northeastern Brazil.Our framework integrating species distribution and connectivity models revealed key habitats and ecological conditions dictating the species distribution and identified main corridors for connecting the remaining populations. Assessing habitat loss across the species range using high-resolution historical land cover data, we found that nearly 6 million hectares of natural vegetation were converted to pasture and agriculture in the past 35 years. In addition, we show that the few known remnant populations of the threebanded armadillo are closely associated with remnant core areas of savanna and grassland, while historical records, reflecting populations now extirpated, were concentrated in areas facing severe habitat loss. Therefore, well-preserved natural vegetation remnants are key to ensuring the species' long-term survival. Based on our multiple findings, we highlight priority areas for resource allocation and conservation actions that can restore connectivity and ultimately benefit other open-dweller species.Our study shows how informed conservation actions can be delineated for threatened species with only limited presence-data records available, a method that can be readily replicated for other poorly known species.