Systematic surveys of massive clumps have been carried out to study the conditions leading to the formation of massive stars. These clumps are typically at large distances and unresolved, so their physical properties cannot be reliably derived from the observations alone. Numerical simulations are needed to interpret the observations. To this end, we generate synthetic Herschel observations using our large-scale star-formation simulation, where massive stars explode as supernovae driving the interstellar-medium turbulence. From the synthetic observations, we compile a catalog of compact sources following the exact same procedure as for the Hi-GAL compact source catalog. We show that the sources from the simulation have observational properties with statistical distributions consistent with the observations. By relating the compact sources from the synthetic observations to their three-dimensional counterparts in the simulation, we find that the synthetic observations overestimate the clump masses by about an order of magnitude on average due to line-of-sight projection, and projection effects are likely to be even worse for Hi-GAL Inner Galaxy sources. We also find that a large fraction of sources classified as protostellar are likely to be starless, and propose a new method to partially discriminate between true and false protostellar sources.