Online marketing is getting more and more popular in today’s digital world. The e-commerce organizations are very cautious about the online reviews provided by customers to enhance their services. In this regard, customer sentiment analysis is highly desirable for organizations. Most of the traditional techniques follow statistical or probability-based mechanisms for customer sentiment analysis. These techniques often suffer from computational accuracy problems. This paper proposes a machine learning-based customer sentiment analysis approach for the betterment of service in online marketing. The proposed approach is designed with a K-nearest neighbor (KNN) classifier, valence-aware dictionary, and reasoning of sentiment techniques. The proposed approach is tested with a real-life customer online review dataset, and the test result shows 92.5% accuracy on the classification of customer reviews.