Summary: Recent empirical evidence suggests that delivery to retail warehouses suffers from a lack of coordination. While carriers try to optimize their routes, they often experience very long waiting times at loading docks, which renders their individual planning useless. To reduce such inefficiencies, carriers need to coordinate. This problem has received considerable attention in practice, but the design of coordination mechanisms is challenging for several reasons: First, the underlying package assignment problem is NP-hard. Second, efficiency, incentive-compatibility, and fairness are important design desiderata, but in most economic environments they are conflicting. Third, the market for logistics services is competitive and price-based mechanisms where carriers might have to pay for time slots suffer from low acceptance. We draw on recent advances in market design, more specifically randomized matching mechanisms, which set incentives for carriers to share information truthfully such that a central entity can coordinate their plans in a fair and approximately efficient way. We use and adapt the existing maximizing cardinal utilities (MAXCU) framework to a retail logistics problem, which yields a new and powerful approach for coordination. We report numerical experiments based on field data from a real-world logistics network to analyze the average reduction in waiting times and the computation times required and compare to first-come, first-served and an auction mechanism. Our results show that randomized matching mechanisms provide an effective means to reduce waiting times at warehouses without requiring monetary transfers by the carriers. They run in polynomial time and provide a practical solution to wide-spread coordination problems.