Over the past few years, clustering-based redshift estimation has emerged as a new way to estimate redshifts and perform extragalactic tomography of arbitrary datasets. On a similar timescale, observations by Planck, WISE, Pan-STARRS and 21cm radio surveys have been used to create a multitude of SFD-type Galactic dust maps. I will explain how clustering-based redshift estimation can be used to test the quality of the seven different dust maps currently available and I will show that extragalactic signatures can be revealed in many of them. When such maps are used for correcting optical magnitudes, we therefore expect biases which are likely to affect the precision of cosmological experiments using supernovae, BAOs, or the growth of structures. I will present possible solutions to alleviate this issue and discuss which map should be used depending on which measurement one wishes to make.