Open Access

Vienna Klein, Claudius Steinhardt

• For providers to stay competitive in a context of continued growth in e-retail sales and increasing customer expectations, same-day delivery options have become very important. Typically, with same-day delivery, customers purchase online and expect to receive their ordered goods within a narrow delivery time span. Providers thus experience substantial operational challenges to run profitable tours and generate sufficiently high contribution margins to cover overhead costs. We address these challenges by combining a demand-management approach with an online tour-planning approach for same-day delivery. More precisely, in order to reserve capacity for high-value customer orders and to guide customer choices toward efficient delivery operations, we propose a demand-management approach that explicitly optimizes the combination of delivery spans and prices which are presented to each incoming customer request. The approach includes an anticipatory sample-scenario based value approximation,For providers to stay competitive in a context of continued growth in e-retail sales and increasing customer expectations, same-day delivery options have become very important. Typically, with same-day delivery, customers purchase online and expect to receive their ordered goods within a narrow delivery time span. Providers thus experience substantial operational challenges to run profitable tours and generate sufficiently high contribution margins to cover overhead costs. We address these challenges by combining a demand-management approach with an online tour-planning approach for same-day delivery. More precisely, in order to reserve capacity for high-value customer orders and to guide customer choices toward efficient delivery operations, we propose a demand-management approach that explicitly optimizes the combination of delivery spans and prices which are presented to each incoming customer request. The approach includes an anticipatory sample-scenario based value approximation, which incorporates a direct online tour-planning heuristic. It does not require extensive offline learning and is scalable to realistically sized instances with multiple vehicles. In a comprehensive computational study, we show that our anticipatory approach can improve the contribution margin by up to 50% compared to a myopic benchmark approach. We also show that solving an explicit pricing optimization problem is a beneficial component of our approach. More precisely, it outperforms both a pure availability control and a simple pricing rule based on opportunity costs. The latter idea is one used in other approaches for related dynamic pricing problems dealt with in the literature.…