To understand the origin of interstellar molecules we rely on astrochemical models, the gas-phase networks of which contain ≥7000 reactions. However, just a tiny fraction of them have parameters derived in laboratory experiments. Theoretical quantum mechanical (QM) calculations can also provide this information. Unfortunately, sometimes theoretical predictions and experimental values disagree, as is the case for the paradigmatic reaction CH₃OH + OH → CH₃O + H₂O. Both laboratory experiments and QM calculations found an unexpected increase in the rate coefficients with decreasing temperature. However, experimental and theoretical estimates of the rate coefficients diverge by up to two orders of magnitude at the low temperatures of interest in interstellar chemistry. This work aims to test whether astronomical observations can help untangle this confusing situation. To this end, we first carried out new QM calculations to derive the rate coefficients of the major destruction reaction of the methoxy radical, CH₃O + H, and then we compared astronomical observations from the IRAM/NOEMA Large Programme SOLIS with astrochemical model predictions. Our new rate coefficient for the CH₃O + H reaction is 5-10 times larger than that in the astrochemical data base KIDA in the 10-100 K range. When including the new methoxy destruction rate coefficients, the comparison between observations and model predictions favours the rate coefficients of the CH₃OH + OH reaction from QM calculations. We conclude that QM calculations are an important alternative to laboratory experiments when it comes to the harsh conditions of interstellar objects and that astronomical observations can be used to constraint the rate coefficients of relevant reactions.