Carbonate clumped isotope thermometry is a useful tool practised in studies of temperature history and fluid composition of surface and subsurface environments, with application to both inorganic and biological precipitates. Its measurements are based upon the propensity with which 13C and 18O isotopes, within a carbonate mineral, are bound to one another, in relation to a stochastic distribution. The quantity of these 13C-18O bonds (commonly referred to as “clumps”) is determined by gas source mass spectrometry on CO2 produced from acid digestion of carbonate minerals, and is controlled by physiochemical parameters of the solution at the time of mineral precipitation. If equilibrium is reached, then 13C-18O abundance, measured against the stochastic distribution and represented by the variable Δ47, can be used to measure the temperature of precipitation of the carbonate without the need to characterize the isotopic composition of coeval fluids. However, long-term reproducibility of these measurements is a critical factor contributing to uncertainties in all calibrations and applications. Here we discuss the impact of using different standardization procedures on the accuracy and precision of Δ47 measurements, as compared across three mass spectrometers with four different configurations within the Tripati Lab at UCLA. Specifically, we assess the long-term reproducibility of carbonate standard Δ47 values across mass spectrometers, using a correction scheme that incorporates either gas and carbonate standards of known composition, or both, and the impact of different approaches for characterizing instrument drift (i.e., averaging for an interval or using a moving window). We also recommend best practices to promote reproducibility.