Optimizing stock allocation in Bol’s complex logistical network

Bol operates five distinct warehouses, each with unique characteristics, capacities, and costs, posing a substantial optimization challenge; where to allocate what product. Our team tackled this complexity by developing a linear optimization model, allocating millions of diverse products to warehouses while considering a large number of constraints. Forecasts on inbound, stock, and outbound levels were being developed and served as crucial inputs for our model.Regularly, we conduct new optimization runs, supplying essential input to continually facilitate allocations through our API within our logistical network.

To enhance decision-making, we created dashboards and automated reports offering insights into optimization runs and forecast outcomes. The project encountered challenges in modeling Bol.’s complex and rapidly changing logistical landscape, as well as engineering hurdles such as reducing runtime and cloud resource usage.

Despite the ongoing nature of this project, we successfully refactored modeling decisions, resulting in improved stock allocations. This, in turn, led to a significant annual reduction in logistical costs. While challenges persist, our continuous efforts underscore the project’s positive impact on Bol’s operational efficiency.