Lazada is one of the biggest marketplaces in Southeast Asia. One of the main features of Lazada is product reviews, any customer who has purchased and used a product from Lazada can provide reviews and ratings on the ones that have been purchased. Sentiment analysis on product reviews can help improve product and service quality, increase customer satisfaction, and improve purchasing decisions. Doing sentiment analysis of product reviews aims to help customers how to feel about the product, reading and analyzing each review manually is very not efficient. Sentiment analysis can automate handling large volumes of data quickly and accurately. In this research, using Lazada product review dataset to analyze sentiment by comparing CNN and Naive Bayes. CNN and Naive Bayes are two common methods used in text analysis and a comparison of their performance can provide the effectiveness of each in analyzing product sentiment. In this study, the authors propose to analyze the sentiment of product reviews using deep learning algorithm with CNN method. The results of this study explain that the CNN method can provide satisfactory results than Naive Bayes. Based on the overall evaluation, CNN gets an accuracy value of 99.31%, precision of 99.31%, recall 99.31%, and f1-score of 99.31%, while Naive Bayes gets the highest accuracy rate of 96.16%, precision 96.34%, recall 96.16%, and f1-score 96.16%.