This study presents a research approach by comparing machine learning models for predicting the lifetime consumption of posts published in brands’ Facebook pages. It turns out the XGBoost model performs the best among all the single models and the ensemble one. Also, feature analysis is done to understand how each of the seven input features influenced the response (category, page total likes, type, month, hour, weekday, paid). The page total likes was considered the most relevant feature for the model, with the largest importance. More- over, categories “Category” and “Type” are consistently important, which is consistent with the empirical study.