Countermeasures to Non- Techincal Losses (NTL) are primarily motivated by the high economic losses they represent. Most solutions focus on detecting suspicious behavior, aiming at identifying fraudulent activities in a data-driven and automated fashion. The direct economic impact associated with these solutions is commonly overlooked. In this work, we focus on the economic impact related to an NTL solution, where the list of customers to be inspected and the inspection routes are optimized. The clients to be examined are selected considering estimates of the fraud probability, the estimated magnitude of the fraud, and the costs associated with a particular inspection route. The mathematical formulation of the optimization problem leverages existing techniques developed in the context of a problem known as “the salesman problem.” We validate our approach in real-world data obtained after inspecting over 168k (fraudulent and typical) customers in Uruguay, South America. The results show the relevance of taking routing costs into account and indicate that the economic harm of NTL can be reduced by 76%.