In the talk, we survey various results on incomplete multiplicative character sums obtained over recent years and some of their applications. The relation with sum-product goes back as far as Vinogradov's shifted product technique and Burgess work. Methods from arithmetic combinatorics and geometry of numbers play a role in extending Burgess results to various multi-dimensional settings but unfortunately did not allow to break the 1/4 threshold so far. We also discuss a unified treatment of the Graham-Ringrose and Postnikov argument for very short character sums when the modulus has only small prime factors and give an application to integral points on Markoff-Hurwitz and Dwork hypersurfaces.