The satellites of the Milky Way and Andromeda form the observational bedrock upon which our understanding of small-scale cosmology rests. In order to understand if this insight generalizes, it is imperative to build a census of faint satellite galaxies in nearby galaxy groups. Intriguingly, to luminosity limits similar to faint classical dwarf galaxies, there is now evidence for a nearly order-of-magnitude range in the number of satellite galaxies of Milky Way-like central galaxies. Furthermore, we find that the number of satellites correlates tightly with the mass of the largest satellite ever accreted by the central galaxy (either already merged or still interacting) — a behavior not seen in state-of-the-art hydrodynamical galaxy formation models. In order to probe this behavior with superior number statistics, sensitivity to ultra-diffuse satellites, and order-of-magnitude fainter luminosity limits, next-generation surveys need to adopt wide-field resolved star techniques. Using deep ground-based Subaru datasets, we present our search methodology and a set of candidate ultra-faint satellites in the M81 group out to galactocentric radii of ~100kpc, complete to M_V~-6.5 (similar to current limits for Andromeda satellites). Star-galaxy separation is the limiting factor of current searches, and we show the dramatic improvements that can be achieved if nearby galaxy groups are surveyed for resolved stars using the Nancy Grace Roman Space Telescope.