Recent years have seen an increase in the deployment of cellular networks in indoor areas. These indoors scenarios are mainly characterized by a high user density as well as fast-changing conditions, which makes them more prone to failures. Furthermore, the substantial development of outdoor and indoor positioning methods will provide a reliable source of location information that is expected to be generally available. Hence, the availability of user positioning is expected to be one of the key enablers to improve the sturdiness and accuracy of automatic failure management and optimization mechanisms. Taking this into consideration, this work proposes a semi-supervised framework for the detection of failures using user positioning.