Uma abordagem meta-heurística para o problema de marketing direto considerando o canibalismo entre produtos

Abstract

In recent years, due to technological advances, it is possible to form considerable databases with complete customer information, facilitating the process of segmentation and individualization in a promotion campaign. This is the basis of direct marketing, known for its customer orientation. Promotional campaigns are fundamental tools of direct marketing to improve a company's economic profit, either by acquiring new customers or by generating additional revenue from existing customers. In the literature, there are several formulations of direct marketing problem. However, given the combinatorial nature of the problem and the high volume of data, the problem is classified as NP-hard, and it is relevant to use heuristic methods to solve it, since it is very difficult to obtain optimal solutions for realistic instances with thousands of customers in a reasonable period of time. In addition, the inclusion of the effect of cannibalism between products was not found in the literature, which occurs when the supply of a product negatively impacts the supply of another similar product from the same company. Therefore, the objective of this work is to develop and evaluate a heuristic algorithm to solve the problem of a promotion campaign considering the cannibalism between products, maximizing customer feedback. The instances used were those available in the literature and were treated statistically, to simulate the effect of cannibalism between products, by cluster analysis. Three heuristic algorithms were proposed, one based on Tabu Search, one on Tabu Search with GRASP, and a third on Tabu Search with GRASP and cannibalism effect. In most of the tested instances the Tabu Search and GRASP algorithm obtained better results than those found in the literature, and the algorithm with cannibalism effect showed worsening in most of the experiments. However, the work proved to be relevant with the results presented, and may help, with the proposed approach, the organizations for optimization and decision making, especially those seeking a competitive advantage with a high efficiency in the allocation of resources.