Retail is changing and even the biggest sector, the food industry, can’t evade itself from this change. Higher requirements, different demands and permanent availability are only some examples of the challenges logistic centres are facing. To fulfill those requirements an excellent logistic, inside and outside warehouses, is needed. A very important part of the logistic processes is the replenishment between buffer stock and overstock. On the one hand costs and used area should be as low as possible and therefore as few pieces as possible of each article should be in buffer stock. On the other hand, it’s mandatory that there are enough pieces, so that no stockouts occur. To react immediately to changes in requirements and avoid stockouts, this thesis is dealing with the classification of articles for an optimized replenishment. Therefore, changes in requirements should be prognosed even before they occur, so that a higher number of articles can be replenished in time. Based on a literature research and data analyses possibilities for classification are evaluated in this thesis. Furthermore, the classification is illustrated in a pseudo-code, which exemplary application on the data already shows some very interesting results.