Poynting flux is the flux of magnetic energy, which is responsible for chromospheric and coronal heating in the solar atmosphere. It is defined as a cross product of the electric and magnetic fields, and in ideal MHD conditions it can be expressed in terms of the magnetic field and plasma velocity. Poynting flux has been computed for active regions and plages, but estimating it in the quiet Sun (QS) remains challenging due to resolution effects and polarimetric noise. However, with the upcoming DKIST capabilities, such estimations will become more feasible than ever before. Here, we study QS Poynting flux in SUNRISE/IMaX observations and MURaM simulations. We explore two methods for inferring transverse velocities from observations—FLCT and a neural network–based method DeepVel—and show DeepVel to be the more suitable method in the context of small-scale QS flows. We investigate the effect of azimuthal ambiguity on Poynting flux estimates, and we describe a new method for azimuth disambiguation. Finally, we use two methods for obtaining the electric field. The first method relies on an idealized Ohm’s law, whereas the second is a state-of-the-art inductive electric field inversion method PDFI_SS. We compare the resulting Poynting flux values with theoretical estimates for chromospheric and coronal energy losses and find that some of the Poynting flux estimates are sufficient to match the losses. Using MURaM simulations, we show that photospheric Poynting fluxes vary significantly with optical depth, and that there is an observational bias that results in underestimated Poynting fluxes due to an unaccounted shear term contribution.