The stellar haloes of galaxies can currently be studied either through observations of resolved halo stars or through surface photometry. Curiously, the two methods appear to give conflicting results, as a number of surface photometry measurements have revealed integrated colours that are too red to be reconciled with the halo properties inferred from the study of resolved stars. Several explanations for this anomaly have been proposed – including dust photoluminescence, extinction of extragalactic background light and a bottom-heavy stellar initial mass function. A decisive test is, however, still lacking. Here, we explain how observations of the halo of a nearby galaxy, involving a combination of both surface photometry and bright star counts, can be used to distinguish between the proposed explanations. We derive the observational requirements for this endeavour and find that star counts in filters VI and surface photometry in filters VIJ appear to be the optimal strategy. Since the required halo star counts are already available for many nearby galaxies, the most challenging part of this test is likely to be the optical surface photometry, which requires several nights of exposure time on a 4–8 m telescope, and the near-infrared surface photometry, which is most readily carried out using the upcoming James Webb Space Telescope.