Abstract The detection of gravitational waves from the binary neuron star merger GW170817 and electromagnetic counterparts GRB170817A and AT2017gfo kick-started the field of gravitational-wave multimessenger astronomy. The optically red to near-infrared emission (“red” component) of AT2017gfo was readily explained as produced by the decay of newly created nuclei produced by rapid neutron capture (a kilonova). However, the ultraviolet to optically blue emission (“blue” component) that was dominant at early times (up to 1.5 days) received no consensus regarding its driving physics. Among many explanations, two leading contenders are kilonova radiation from a lanthanide-poor ejecta component and shock interaction (cocoon emission). In this work, we simulate AT2017gfo-like light curves and perform a Bayesian analysis to study whether an ultraviolet satellite capable of rapid gravitational-wave follow-up, could distinguish between physical processes driving the early “blue” component. We find that ultraviolet data starting at 1.2 hr distinguishes the two early radiation models up to 160 Mpc, implying that an ultraviolet mission like Dorado would significantly contribute to insights into the driving emission physics of the postmerger system. While the same ultraviolet data and optical data starting at 12 hr have limited ability to constrain model parameters separately, the combination of the two unlocks tight constraints for all but one parameter of the kilonova model up to 160 Mpc. We further find that a Dorado-like ultraviolet satellite can distinguish the early radiation models up to at least 130 (60) Mpc if data collection starts within 3.2 (5.2) hr for AT2017gfo-like light curves.