The retail markets are creating loads of transactional data but the tra-ditional methods of segmentation cannot keep up with the emergent patterns of consumer behavior. This work proposes a data-driven model which presents retail market as a complex system, and reveals some unseen relations between product sales and customer buying behavior. Using a massive transaction dataset of supermarkets, we are postulating a Purchase Function that is able to predict consumer demands systematically, hierarchies of products and refines targeted marketing strategies. Our solution offers practical recommendations on individualized promotions, demand prediction, and store op-benefit, and offers an alternative view of how data science can be used to drive com-competitive advantages in retail marketing. Moreover, the offered framework shows that it can be slightly adjusted to various retail settings and markets, which implies that it can be widely applicable to data-based personalization approaches in the markets outside the examined one in Italy.