Humans are highly subjective beings and present their opinions on products and services in the form of reviews, which need to be analyzed to understand the consumer base. However, deciphering the consumer sentiment from online reviews is a massive task due to the diversity in the reviews and their unstructured nature. To tackle this challenge, in this study we employ the BERT (Bidirectional Encoder Representations from Transformers) model to predict numerical ratings from textual reviews from a Flipkart product review dataset sourced from Kaggle. Our findings demonstrate BERT’s suitability for analyzing diverse and complex textual data, with an impressive accuracy of 73%, due to its ability to employ contextual analysis. This study showcases the potential of utilizing advanced NLP techniques to derive insights from otherwise huge and undecipherable data, which can be used to make well informed decisions by businesses looking to maximize gains and customers looking for make better purchasing choices alike.