TikTok has experienced significant growth, capturing the interest of academics and practitioners aiming to optimise content and marketing strategies. In the political sphere, analysing TikTok trends is essential for understanding public opinion and improving campaign effectiveness. This study introduces a predictive model designed to identify the factors influencing the virality of political videos during Ecuador’s 2023 presidential runoff election between Daniel Noboa and Luisa González. Machine learning algorithms were employed to evaluate variables such as the day of publication, video duration, hashtags, topics, sentiment in descriptions, and types of sound. Performance was assessed using metrics including precision, recall, F1-score, and area under the curve (AUC). For Noboa, the k-nearest neighbours (KNN) model achieved the highest predictive accuracy, with a precision of 0.625, recall of 0.833, and F1-score of 0.714, while the artificial neural network (ANN) model recorded the highest AUC at 0.806. For González, ANN emerged as the most effective model, achieving balanced performance across all metrics with a precision, recall, and F1-score of 0.714, and an AUC of 0.786. The results revealed that the day of publication was the most influential factor for Noboa’s videos, whereas the topic of the videos was most significant for González. This study contributes to a greater understanding of the dynamics of political content on social platforms, suggesting that publication timing and thematic content are critical for maximising video reach in political campaigns, while video duration and description sentiment exert less influence.