In the era of advanced digital commerce, consumers’ perceptions of products are increasingly influenced by customer reviews, which also significantly impact their purchasing decisions. This study provides an in-depth sentiment analysis and trends in Amazon product reviews. The study conducts a comprehensive investigation focused on Alexa terminals, utilizing supervised Machine Learning (ML) algorithms based on the Cross-Industry Standard Process for Data Mining (CRISP-DM) architecture. Four diverse ML classifiers: Random Forest (RF), Modified Decision Forest (MDF), Extreme Gradient Boosting (XGBoost), and K-Nearest Neighbors (KNN) were utilized to model and classify the sentiment of the customers effectively. A benchmark Amazon Alexa Reviews dataset, which included a rich textual description of customer opinions, was utilized for implementing and evaluating the proposed work. Google Colab with Python is used to conduct experiments that achieve scalability and reproducibility. Based on standard classification metrics, the assessment of the produced models utilized different data split configurations: (30–70, 40–60, 50–50, 60–40, and 70–30) for training and testing, respectively. Experimental results reveal that RF outperforms the other algorithms across all different splits in terms of precision, accuracy, and recall.