Energy-resolving gamma-ray detectors are of particular interest for the detection of illicit radioactive materials at border crossings and other portals because they offer fast, contactless screening that can discriminate between dangerous and benign materials. Among detector classes, scintillators offer an intriguing balance between cost and performance, but current technologies rely on single-crystal materials that are not scalable to portal-relevant detector sizes. Thus, there is a recognized need for novel, processible, high-performance scintillating materials or composites. Composites based on semiconductor nanocrystal quantum dots (QDs) are of interest because of their potentially high gamma-stopping power, high emission quantum yields, and low-cost solution synthesis and processing. Yet the performance of these and other granular nanomaterials has not met expectations. We suggest that this is due to the general lack of insight into the gamma-to-photons transduction process within these inherently more complex materials, which reduces the development and refinement of candidates to simple trial-and-error. Here, we describe the development of ultrafast transient cathodoluminescence as a unique spectroscopic tool for probing the population of excited states formed within a material during scintillation, and thus determining the major sources of energy loss. Our analysis shows that in the case of CdSe/ZnS core/shell QDs, any efficiency loss due to previously blamed factors of low-stopping power and high reabsorptive losses are likely dwarfed by the losses attributable to efficient, non-radiative Auger recombination. We examine how we reached this conclusion, and how this insight defines the characteristics needed in the next generation of scintillating QD composites.