There is currently a gap in satellite observations of the moisture structure during heavy precipitation conditions, since infrared and microwave sounders cannot sense water‐vapour structure near the surface in the presence of intense precipitation. Conversely, Global Navigation Satellite System (GNSS) radio occultations (RO) can profile the moisture structure with high precision and vertical resolution, but cannot indicate the presence of precipitation directly. Polarimetric RO (PRO) measurements have been proposed as a method to characterize heavy rain in GNSS RO, by measuring the polarimetric differential phase delay induced by large size hydrometeors. Previous studies have shown that the PRO polarimetric phase shift is sensitive to the path‐integrated rain rate under intense precipitation scenarios, but there is no current method to invert PRO measurements into quantitative estimates of the path‐averaged rain rate. In this manuscript, a probabilistic inversion approach to the GNSS PRO observables is proposed, where the GPM precipitation products are used for the construction of an a priori look‐up table (LUT) database. The performance of the LUTs is assessed for use in the inversion of satellite‐based GNSS PRO observations, based on synthetically generated PRO data of actual events, which correspond to co‐locations between GNSS RO profiles and the TRMM observations. The synthetic data include end‐to‐end propagation effects of the polarimetric observables and a simple separation algorithm to isolate the hydrometeor component of the observation. The assessment results in agreement better than ±1 mm/hr between the reference LUT and the actual rain statistics of the synthetic data, proving the suitability of the GPM‐based probabilistic inversion tool. These findings indicate that the GNSS PRO products are capable of extending the current GNSS RO ones by associating indications of rain‐rate probabilities at different altitudes, at ∼250 m vertical resolution and under intense precipitation scenarios with the standard vertical thermodynamic profiles.