The generation of dark matter in late decays of quasi-stable massive particles has been proposed as a viable framework to address the excess of power found in numerical N-body simulations for cold dark matter cosmologies. We identify a convenient set of variables to illustrate which requirements need to be satisfied in any generic particle physics model to address the small scale problems and to fulfill other astrophysical constraints. As a result of this model-independent analysis, we point out that meeting these requirements in a completely natural way is inherently difficult. In particular, we re-examine the role of gravitinos and Kaluza-Klein gravitons in this context and find them disfavoured as a solution to the small-scale problems in case they are DM candidates generated in the decay of thermally produced WIMPs. We propose right-handed sneutrinos and right-handed Kaluza-Klein neutrinos as alternatives. We find that they are viable dark matter candidates, but that they can contribute to a solution of the small scale problems only in case the associated Dirac neutrino mass term appears as a subdominant contribution in the neutrino mass matrix.