Heuristic optimization algorithms for advertising campaigns

In this paper, two optimization problems within the scope of marketing campaign design are studied. In particular, two ad positioning problems are analyzed with the objective of minimizing the cost of all the chosen media while ensuring that a set of constraints is fulfilled. In the first problem, a given minimum number of impressions (views) in each population segment is required to be reached as a constraint. In the second problem, the constraints involve achieving, in each population segment, a given minimum probability that any individual will see the ad at least once. In this second case, media dependencies are defined for each population segment: independence, exclusion, and inclusion. Since both problems are Log-APX-hard, heuristic methods are used to solve them. More specifically, greedy and genetic algorithms, as well as particle swarm optimization, are applied to address these problems. Furthermore, the usefulness of these algorithms is evaluated through concrete case studies using real data on ad prices and views. • Formally specifying two problems in the context of media planning. • Providing heuristic solutions for them by means of Genetic Algorithms. • Evaluating our implementations to deal with two concrete case studies.