General relativity as a classical field theory does not predict gravitationally induced entanglement, as such, recent proposals seek an empirical demonstration of this feature which would represent a significant milestone for physics. We introduce improvements to a spin witness protocol that reduce the highly challenging experimental requirements. After rigorously assessing approximations from the original proposal [S. Bose et al. Phys. Rev. Lett. 119, 240401 (2017)], we focus on entanglement witnessing. We propose a new witness which greatly reduces the required interaction time, thereby making the experiment feasible for higher decoherence rates, and we show how statistical analysis can separate the gravitational contribution from other possibly dominant and ill-known interactions. We point out a potential loophole and show how it can be closed using state tomography.
