NLP Based Sentiment Analysis For Understanding Consumer Experience In Omnichannel Retail

Abstract

This study aims to analyze consumer experience in omnichannel retail through a Natural Language Processing (NLP)-based sentiment analysis approach. Six problem formulations were addressed through a survey of 500 respondents in Indonesia who use omnichannel retail applications, comprising 8 main applications. The results of the study show: (1) consumer experience patterns are reflected in the distribution of sentiments that vary in the order of Product (3.78) > Application (3.73) > Delivery (3.65) > Customer Service (3.61); (2) four main factors are identified through topic modeling: delivery (28%), application (26%), customer service (22%), and product (24%); (3) all three dimensions of channel integration quality have a positive effect on consumer sentiment, with integrated service quality having the strongest effect (r = 0.62, p < 0.001), followed by consistency (r = 0.59) and ease of switching (r = 0.58); (4) there is a significant difference in the distribution of sentiments between service aspects (F = 4.892, p < 0.01), with product as the main strength and customer service as the biggest weakness; (5) age and education influence consumer perception, while gender, occupation, and income do not show significant differences; (6) star ratings are highly correlated with consumer satisfaction (r = 0.85, p < 0.001), validating its use as a reliable indicator of sentiment. This research contributes to the development of omnichannel retail literature in developing countries. It offers practical implications for retail managers to prioritize improvements in customer service, ease of returns, and delivery speed.