In multi-stage production structures, particularly in network-based industries, optimal allocation of shared resources plays a critical role in enhancing the efficiency and operational coherence of the entire system. This study proposes an innovative and generalized model of Data Envelopment Analysis (DEA) in the form of a Centralized Inverse Network DEA. The model utilizes the Maximum Slack-Based Inefficiency (MSBI) index to prescribe optimal values for inputs, intermediate products, and outputs across a network of Decision-Making Units (DMUs). The strength of the proposed model lies in its simultaneous integration of network structure, prescriptive DEA approach, and Centralized Resource Allocation (CRA) constraints, making it a powerful tool for redistributing shared resources within large-scale organizations and holdings. The model is capable of identifying inefficiency bottlenecks, analyzing managerial scenarios, reallocating resources, and optimizing the overall system performance. For empirical validation, the model was implemented on a real network of 16 cement factories under a large industrial holding. The results demonstrated that the application of the proposed model led to a significant increase in average efficiency, a notable reduction in the standard deviation of inefficiencies, and a substantial improvement in meeting production targets. The developed model is extendable to other industries with similar network structures (such as steel, petrochemical, and pharmaceutical industries) and can serve as an effective decision-support tool in strategic management and organizational productivity policymaking.