In cosmological analyses that measure lensing by large-scale structure, knowledge of the underlying redshift distribution of the galaxies is key. A full complement of spectroscopic measurements is expensive, and it is common to use lower resolution photometric information to constrain redshift distributions. However, these estimates can be biased at a level on par with the statistical precision of current lensing data sets. I will discuss how to leverage spatially overlapping CMB and spectroscopic (not necessarily representative of the photometric sample selection) surveys to control this photometric redshift bias and mitigate its effect on output cosmological parameters.