Electromagnetic induction (EMI) methods are often used to map rapidly large areas with minimal logistical efforts. However, they are limited by a small number of frequencies and by their severe ill-posedness. On the other hand, electrical resistivity tomography (ERT) results are generally considered more reliable, with no need for specific calibration procedures and easy 2D/3D inversion. Still, ERT surveys are definitely more time-consuming, and, ideally, an approach with the advantages of both EMI and ERT would be optimal. The present research addresses this issue by incorporating realistic constraints into EMI inversion, going beyond simplistic spatial constraints like smooth or sharp regularization terms, while taking into consideration the ancillary information already available about the investigated site. We demonstrate how additional pre-existing information, such as a reference model (i.e., an existing ERT section) can enhance the EMI inversion. The study verifies the results against observations from boreholes.