Customer loyalty depends on a combination of objective and subjective factors. For retailers, the benefit having a positive perception in the public is the credibility transmitted to their brands. This investigation is focused on the implementation of artificial neural networks (ANN) to predict customers’ loyalty to supermarket chains measured in terms of their purchase frequency. The research is carried out following a classic 4-stage methodology (analysis, design, development, and validation). During the analysis, the results of a customer loyalty survey are preprocessed. During the design, the full questionnaire is divided into several domains (e.g. products’ quality, buying experience, etc.). Each domain is then used to build and compare predictive models. The original dataset containing 1.050 surveys is split up in two sets: 80% of data for training and test, and the remaining 20% for validation. To predict customer loyalty twelve different ANN-based models are built. The results reveal that the proposed ANN-based models can predict correctly 60% and 72% of customers’ purchase frequency. In conclusion, predictive ANN-based models help determine how often an average customer made purchases on a given supermarket chain with a reasonable degree of certainty by means of analyzing some of their habits and preferences.