The objective was to integrate machine learning models like Long Short-Term Memory (LSTM), Artificial Neural Networks (ANNs), Random Forest (RF), and Decision Trees (DT), into performing sentiment analysis on consumers’ reviews on e-commerce sites, on real-time. Consumer review of 2000 samples were gathered and distributed evenly on positive and negative sentiments and then preprocessed to remove noise and normalize text values data. The models trained and tested on most influential metrics like precision, recall, F1-score, accuracy, and AUC-ROC. The LSTM model, too, exhibited best performance among the best-performing models, especially the best precision (92.0%), recall (90.0%), F1-score (91.0%), and accuracy (91.5%), and an AUC-ROC measuring 96.0%. The second was the ANN model, accuracy 87.0%, recall 85.0%, F1-score 86.0%, and accuracy 86.8%, and an AUC-ROC measuring 91.0%. Random Forest model won an accuracy 83.0%, recall 81.0%, F1-score 82.0%, accuracy 82.7%, and an AUC-ROC measuring 88.0%, and, however, worst among them was experienced by the Decision Tree model gaining precision 79.0%, recall 76.0%, F1-score 77.0%, accuracy 78.5%, and an AUC-ROC measuring 83.0%. These results show the advantage of deep learning models, especially LSTM, to detect the subtleness on consumers’ sentiments over machine learning models. The findings show the prospectivity of deep learning to enhance the precision on sentiment analysis, and thus customer experience and decision-making within e-commerce industry settings.