Not yet reviewed by Pith; the record is open.

Long-term quantum coherence constitutes one of the main challenges when engineering quantum devices. However, easily accessible means to quantify complex decoherence mechanisms are not readily available, nor are sufficiently stable systems. We harness novel phase-space methods - expressed through non-Gaussian convolutions of highly singular Glauber-Sudarshan quasiprobabilities - to dynamically monitor quantum coherence in polariton condensates with significantly enhanced coherence times. Via intensity- and time-resolved reconstructions of such phase-space functions from homodyne detection data, we probe the systems's resourcefulness for quantum information processing up to the nanosecond regime. Our experimental findings are confirmed through numerical simulations for which we develop an approach that renders established algorithms compatible with our methodology. In contrast to commonly applied phase-space functions, our distributions can be directly sampled from measured data, including uncertainties, and yield a simple operational measure of quantum coherence via the distribution's variance in phase. Therefore, we present a broadly applicable framework and a platform to explore time-dependent quantum phenomena and resources.