Recent research publications in the polarization literature have discussed methods of correcting for azimuthal alignment errors of optical elements in postprocessing. However, we show that high angular precision is not difficult to achieve during system alignment, so that postprocessing correction should be unnecessary. We estimate the alignment precision achievable for linear polarizers and waveplates in polarization systems. This shows that using an optical signal model for alignment allows a precision limited by the quality of the optics and detectors rather than the quality of the mechanics, rendering millidegree alignment precision possible with ordinary rotational mounts.