Case study: Optimizing initial distribution

Especially for chain stores in the fashion industry, it is important to ensure that new products are available in each store in the sizes that are also in demand by customers. Most retailers use complex manual procedures to calculate the demand along with the associated inaccuracies. For a fashion retailer with more than 200 stores, COSMO CONSULT developed an optimization model that distributes quantities and sizes ideally and automatically to the individual stores. This made it possible to supply each individual store according to the respective demand, reduce residual stocks and significantly increase sales.

• Supplying the fashion retail industry according to the demand 
• Optimizing demand based on past sales
• Increasing margins and reducing residual stocks

The demand-based size distribution of products to individual stores has a large impact on the margin and is an ongoing challenge faced by fashion retailers.

The quantity and size distribution of a fashion retailer with around 200 stores had previously largely been carried out manually. For this purpose, the individual stores were divided into different classes, for example according store size. Each class was then assigned a predefined size key for each individual product group. This was then used to determine the delivery quantity of the new range of products.

However, there was no further differentiation between the different stores of a class. Also, changes in the shopping behaviors were not regularly analyzed.

COSMO CONSULT was commissioned to develop an automated algorithm that can efficiently determine the initial distribution to the individual stores under given constraints through intelligent data analysis and optimization algorithms.

The first step was to accurately analyze the sales and inventory data of the recent past for each product group. Thus, the project team considered the selling speed of each individual size of a product with respect to the other sizes. The result was implemented into a mixed-integer optimization model, which determined the profit-optimized distribution of the total quantity per store and size. Other restrictions were taken into account such as the minimum quantities per size and store and the delivery lots. These steps are now automated with regular runs.

By analyzing and optimizing each store as well as the sales history and inventory data of each size, it was possible to supply each individual store according to the demand. At the same time, residual stocks were reduced and sales were significantly increased. The optimization process also takes into account more complex relationships when it comes to including a limited number of any lot types in the calculation. Due to the high degree of automation and the short calculation time, the solution can also be used at short notice to react optimally and quickly to order deviations for new deliveries.