The distribution of aerosols in the atmosphere above cosmic ray fluorescence detectors must be well characterised in order to precisely recover extensive air shower properties such as the calorimetric energy, $E$, and depth of shower maximum, $X_\mathrm{max}$. The Pierre Auger Observatory uses two centrally located laser facilities to measure the vertical aerosol optical depth profile (VAOD) every hour. It is assumed that the night with the clearest atmosphere each year is effectively aerosol free and that it is an appropriate reference to set the absolute scale of VAOD throughout that year. We review the successes of this method and its associated sources of systematic uncertainty, then present a new cross-check of measured VAOD using air shower events observed in stereo mode. Special attention is paid to quantifying the uncertainties on this result. As the technique is only sensitive to VAOD bias at a fixed altitude, we combine it with a study of aerosol profiles independently measured using a less-sensitive Raman lidar system. This allows us to derive a complete model of the upper limit on the possible bias in the average measured VAOD, which we attribute primarily to an uncertainty on whether the annual reference nights are completely aerosol free. We formulate a correction for this bias and apply it retroactively to all VAOD measurements, then repeat the analysis of the complete air shower dataset and discuss the small but significant effect of this new correction on $E$ and $X_\mathrm{max}$. This correction is now fully integrated into the Auger analysis chain.