The availability of large data sets with stellar distance and polarization information will enable a tomographic reconstruction of the (plane-of-the-sky-projected) interstellar magnetic field in the near future. We demonstrate the feasibility of such a decomposition within a small region of the diffuse interstellar medium (ISM). We combine measurements of starlight (R-band) linear polarization obtained using the RoboPol polarimeter with stellar distances from the second Gaia data release. The stellar sample is brighter than 17 mag in the R-band and reaches out to several kiloparsecs from the Sun. H i emission spectra reveal the existence of two distinct clouds along the line of sight. We decompose the line-of-sight-integrated stellar polarizations to obtain the mean polarization properties of the two clouds. The two clouds exhibit significant differences in terms of column density and polarization properties. Their mean plane-of-the-sky magnetic field orientation differs by 60°. We show how our tomographic decomposition can be used to constrain our estimates of the polarizing efficiency of the clouds as well as the frequency dependence of the polarization angle of polarized dust emission. We also demonstrate a new method to constrain cloud distances based on this decomposition. Our results represent a preview of the wealth of information that can be obtained from a tomographic map of the ISM magnetic field.