Empty Container Repositioning (ECR) remains one of the most persistent and costly inefficiencies in global container logistics. To balance asymmetric trade flows at global, regional, and local scales, empty containers must be relocated from surplus locations to areas facing deficits. These movements typically generate no revenue for shipping lines. While the ECR problem has been widely studied at the global level, regional and local repositioning remain underexplored. This study investigates key barriers preventing direct inland repositioning between surplus and deficit locations within Maersk's hinterland networks of Rotterdam and Antwerp. A multi-method research approach was adopted, encompassing network analysis using internal Maersk data, stakeholder identification through semi-structured interviews, operational feasibility assessments, and financial analysis of repositioning corridors. Only 12% of container movements currently use direct repositioning, underscoring an overreliance on hub-based repositioning. Critical barriers were identified at both depot and corridor levels. Depot-level barriers include limited inter-vendor collaboration, fluctuating stock levels, container quality requirements, and seasonal flow variations. Corridor-level barriers encompass off-route positioning, fixed transportation schedules, lack of appropriate services, and physical constraints such as barge sizes. Financial assessments demonstrate clear cost advantages for direct repositioning by avoiding additional handling steps associated with hub-based movements. Operational insights highlight the impact of stakeholder interactions, coordination constraints, and planning routines on repositioning efficiency. These findings underscore the need for better stakeholder coordination and more flexible planning routines to make direct inland repositioning viable. By framing ECR as a coordination problem rather than purely an optimization task, this study offers practical insights for improving inland container flows in complex multi-actor networks.