In multi-variant series production, companies face the challenge of ensuring factors like flexibility and efficiency in planning and ordering despite external and internal turbulences, while simultaneously guaranteeing production as well as procurement stability. To address this challenge, the concept of planned orders has the potential to holistically optimize over the entire planning and ordering process from generating fully specified product configurations, followed by the scheduling of these to valid virtual production programs, until the assignment of incoming customer and stock orders. In doing so, precise material requirements forecasting and simulation-based analyses of future production scenarios are directly integrated in the order-to-delivery (OTD) process, facilitating the early identification and avoidance of potential bottlenecks and risks. This paper introduces a data-driven approach that optimizes the distribution of build-to-stock (BTS) orders across the global sales network, enabling the recommendation of demand-oriented product configurations to dealers. Therefore, several data sources such as historical orders, product structure and market as well as dealer information are taken into account to train the developed algorithm. Coupled with the consideration of existing constraints and given uncertainties, the method offers the ability to allocate pre-planned orders across the dealer network, with the objective to align future customer demand optimally. The presented approach is validated by a real-world use case of the Dr. Ing. h.c. F. Porsche AG (PAG) to demonstrate the potential of significantly improving the allocation of orders to the dealers in terms of fulfilling upcoming customer requirements, while guaranteeing stable production and procurement processes.