The origin and properties of black hole seeds that grow to produce the detected population of supermassive black holes are unconstrained at present. Despite the existence of several potentially feasible channels for the production of initial seeds in the high redshift universe, since even actively growing seeds are not directly observable at these epochs, discriminating between models remains challenging. Several new observables that encapsulate information about seeding have been proposed in recent years, and these offer exciting prospects for truly unraveling the nature of black hole seeds in the coming years. One of the key challenges for this task lies in the complexity of the problem, the required disentangling of the confounding effects of accretion physics and mergers, as mergers and accretion events over cosmic time stand to erase these initial conditions. Nevertheless, some unique signatures of seeding do survive and still exist in: local scaling relations between black holes and their galaxy hosts at low-masses; in high-redshift luminosity functions of accreting black holes; and in the total number and mass functions of gravitational wave coalescence events from merging binary black holes. One of the clearest discriminants for seed models are these high redshift gravitational wave detections of mergers from space detectable in the milliHertz range. These predicted event rates offer the most direct constraints on the properties of initial black hole seeds. Improving our theoretical understanding of black hole dynamics and accretion will also be pivotal in constraining seeding models in combination with the wide range of multi-messenger data.