Identifying customer preferences on two competitive startupproducts: An analysis of sentiment expressions and textmining from Twitter data

Abstract

Startups have great potential to grow and scale up their business quickly; moreover, they have an essential role in the growth of the country and the global economy. However, with the high risk of failure, startup success needs to be supported and concerned. The success of startups depends on market needs and expectations, which are currently highly uncertain, dynamic, and chaotic. Thus, it is necessary to identify and monitor customer preferences for startup products/services. This research identifies the customer preferences of two competitive food delivery startups that have been successful, namely Go Food and Grab Food. With increasing customer opinions on social media, Twitter data can be used to explore customer needs and preferences. However, social media data like Twitter tend to be unstructured, informal, and noisy, so data mining mechanisms are needed. Using sentiment analysis and text mining methods, this study explores and compares customer preferences for successful startup products, which has yet to be done in previous studies. The sentiment analysis results show the dominance of positive customer opinions and expressions of the products/services offered. Furthermore, customer product aspects reviewed positively and negatively by customers were analyzed more deeply using text mining to find the strength and weaknesses of these two businesses. The method and analysis of this paper help monitor customer opinions in real-time, both related to their satisfaction and complaints. Finally, the research results have been validated by comparing sentiment analysis classifications using machine learning and manual analysis by experts, which show an accuracy of 85% and 86% in Go Food and Grab Food reviews.

Keywords

• startups
• food delivery services
• sentiment analysis
• text mining
• customer preference
• customer satisfaction