Diverse socioeconomic impacts from power outages in distribution networks are often caused by failures, and estimating their location frequently requires dispatching crews. However, in-field crews can reduce outage areas by network reconfiguration using manually controlled switches. In previous studies, a preset patrolling strategy disregarding the fact that several layers of information are interdependent led to suboptimal recovery of reliability indices, despite the fact that there was eventual service restoration. Thus, this article introduces a bi-level optimization method in which its upper level identifies the number and sequence of restoration stages through switching operations, and its lower level maps out the fastest crew routes from road networks. In addition, Monte Carlo simulations followed by Bayesian models were performed to identify limitations and put forward prepositioning suggestions for crews based on community detection. Their results show that the proposed method is well capable of coupling patrolling procedures with service recovery processes by merging diverse layers of information. Bayesian models were well capable of calculating tradeoffs for several manual recovery tactics, in addition to reflecting the impact of road network topics on power distribution reliability. Ultimately, the suggested method proved suitable to reduce reliability indices by merging several information layers, given that patrolling strategies were optimized and service was restored efficiently.