Sentiment analysis of product reviews is becoming more popular due to the potential it has to enhance e-commerce services and product quality. Deep Learning (DL), Machine Learning (ML), and natural language processing (NLP) approaches are used to classify the sentiment of product evaluations. Customer reviews have an impact on purchasing decisions, which is why sentiment research is important for online retailers. This study investigates the sentiment categorization of Amazon product evaluations using several ML algorithms, one of which being Convolutional Neural Networks (CNNs). The dataset, sourced from Kaggle, comprises over 34,000 reviews spanning various product categories. The BoW method is used for feature extraction after text preparation, which includes cleaning, tokenization, and stop-word removal. The CNN model is trained with optimized hyperparameters, achieving an accuracy of 94%, precision of 94%, F1score of 99%, and recall of 96.03%. Comparative analysis with Logistic Regression (LR), Random Forest (RF), Multinomial Naïve Bayes (MNB), and Naïve Bayes (NB) demonstrates CNN’s superior performance in sentiment classification. Experimental results highlight CNN’s effectiveness in capturing sentiment nuances, reducing misclassification, and improving automated sentiment analysis for e-commerce applications.