While new, advanced, ground-based instrumentation continues to produce new exoplanet discoveries and provide further insights into exoplanet formation and evolution, our desire to discover and characterize planets of Earth size about stars of all types and ages necessitates dedicated, imaging space instruments. Given the high costs and complexities of space observatories, it is vital to build confidence in a proposed instrument’s capabilities during its design phase, and much effort has been devoted to predicting the performance of various flavors of space- based exoplanet imagers. The fundamental problem with trying to answer the question of how many exoplanets a given instrument will discover is that the number of discoverable planets is unknown, and so all results are entirely dependent on the assumptions made about the population of planets being studied. Here, we explore an alternate approach, which involves explicitly separating instrumental and mission biasing from the assumptions made about planet distributions. This allows us to calculate a mission’s ‘depth of search’-a metric independent of the planetary population and defined as the fraction of the contrast–projected separation space reached by a given instrument for a fixed planetary radius and semi-major axis. When multiplied by an assumed occurrence rate for planets at this radius and semi-major axis (derived from an assumed planetary population), this yields the expected number of detections by the instrument for that population. Integrating over the full ranges of semi-major axis and planetary radius provides estimates of planet yield for a full mission. We use this metric to evaluate the coronagraphs under development for the WFIRST mission under different operating assumptions. We also compare the results of convolving the depth of search with an assumed planetary population to those derived by running full mission simulations based on that same population.