Top-flop division is based on the idea that the demand percentages of the 'top' and the 'flop' SKUs in a group of SKUs are fairly stable over time. For example, the 33% best-selling SKUs in a product group of t-shirts represent about 70% of total t-shirt demand, while the 33% worst-selling t-shirts only represent 5% (and the remaining 33% of SKUs represent 25%). Note that the number of top-to-flop categories is 3 in this example, but can be any number in general.

We remark that this bears similarities with the classiflcation of SKUs in ABC systems, which are common in practice. The idea has also been suggested by experts from the case study company. This method is included because it is new and intuitively attractive.

So, the historical SKUs in H are divided into a set C equally sized categories, ranging from top (highest total demand) to flop (lowest total demand). These categories are numbered 1, ... ,C and the corresponding subsets of H are denoted by H1; ... ;HC.If H/C is integer, then each category contains H/C SKUs. Otherwise, categories1, ... ,H bH/CcC contain dH/Ce SKUs and the remaining categories contain bH/Cc SKUs. Here, bxc denotes x rounded down to the nearest integer and dxe denotes x rounded up to the nearest integer. Let Hc, c =1, ... ,C, denote the number of SKUs in Hc .