We are now routinely detecting gravitational waves (GW) emitted by merging black holes and neutron stars. Those are the afterlives of massive stars that formed all across the Universe -at different cosmic times and with different metallicities (abundances of elements heavier than helium). Birth metallicity plays an important role in the evolution of massive stars. Consequently, the population properties of mergers are sensitive to the metallicity dependent cosmic star formation history (fSFR(Z,z)). In particular, within the isolated formation scenarios (the focus of this paper), a strong low metallicity preference of the formation of mergers involving black holes was found. The origin of this dependence and its consequences are discussed. Most importantly, uncertainty in the fSFR(Z,z) (substantial even at low redshifts, especially at low metallicity) cannot be ignored in the models. This poses a significant challenge for the interpretation of the observed GW source population properties. Possible paths for improvements and the role of future GW detectors are considered. Recent efforts to determine fSFR(Z,z) and the factors that dominate its uncertainty are summarized. Many of those factors are related to the properties of galaxies that are faint and distant and therefore difficult to observe. The fact that they leave imprint on the properties of mergers as a function of cosmic time makes future GW observations a promising (and complementary to electromagnetic observations) tool to study chemical evolution of galaxies.