Lepton flavor violating processes are optimal observables to test new physics, since they are forbidden in the Standard Model while they may be generated in new theories. The usual approach to these processes is to perform the computations in the physical basis; nevertheless this may lose track of the dependence on some of the fundamental parameters, in particular on those at the origin of the flavor violation. Consequently, in order to obtain analytical expressions directly in terms of these parameters, flavor techniques are often preferred. In this work, we focus on the mass insertion approximation technique, which works with the interaction states instead of the physical ones, and provides diagrammatic expansions of the observables. After reviewing the basics of this technique with two simple examples, we apply it to the lepton flavor violating Higgs decays in the framework of a general type-I seesaw model with an arbitrary number of right-handed neutrinos. We derive an effective vertex valid to compute these observables when the right-handed neutrino masses are above the electroweak scale and show that we recover previous results obtained for low scale seesaws. Finally, we apply current constraints on the model to conclude on maximum Higgs decay rates, which unfortunately are far from current experimental sensitivities.