<p>Data science in the dairy industry includes studies on animals, e.g., disease detection, and farms, e.g., milk production forecasting. Most studies are based on statistics and machine learning and do not leverage the advantages of operations research in solving multicriteria problems that incorporate both quantitative and qualitative information. This research proposes a data-driven decision-making framework to address a relatively underexplored problem in the dairy industry: determining the optimal location for a dairy factory. This framework presents four points of originality. The attractiveness of a location is influenced by its proximity to other locations. Factory installation attractiveness encompasses multiple criteria aggregated into one-dimensional measures known as composite indicators. The explanatory and informative power, the proportion of outliers, and the classification uncertainty serve to select composite indicators with sufficient quality to be combined in the composite indicator of attractiveness for the installation of a dairy factory. Decision makers participate in defining the factory’s installation site by interpreting potentially contradictory results. The results of applying this framework reveal that the Brazilian states with the highest attractiveness have more factories than expected, suggesting that the composite indicator of attractiveness is consistent with reality, as many companies choose these states to set up their plants. However, the existence of more factories than expected suggests that these states have an excess of factories and that competition among these plants is more intense. The third most attractive state for setting up a plant has fewer plants than expected, indicating that the state is attractive in terms of the criteria and competition.</p>