CHAT BASED SERVQUAL MEASUREMENT USING TEXT CLASSIFICATION (CASE STUDY: PAPERLUST)

• Main Author: APRIANDITO ARYA SAPUTRA, MUHAMMAD
• Format: Theses
• Language: Indonesia
• Subjects: Manajemen umum
• Online Access: https://digilib.itb.ac.id/gdl/view/53147
• Institution: Institut Teknologi Bandung

The internet’s presence allows the purchasing and selling process to be carried out online via the internet or Electronic Commerce / Ecommerce. E-commerce offers substantial market opportunities and advantages but also presents significant challenges. A survey states that 90% of e-commerce fail within the first 120 days. One of the reasons for the failure of E-commerce is low service quality. Quality of service is crucial for the success of any business. Good service quality creates a competitive advantage for the company, while low service quality can impact customer switching to competitor services. Paperlust, a company engaged in online design and printing, faces a decline in purchase levels throughout 2019-2020. Paperlust has not taken measurements and evaluations related to the quality of services provided so the unsatisfying aspects of service is still unknown. This finding is in line with research that proves that service quality significantly influences consumer purchasing decisions. Thus, measuring satisfaction in every aspect of service quality is essential to do. In this study, the authors propose using text mining methods to measure service quality based on customer conversation data. The text mining technique used is text classification with Multi Layer Perceptron Artificial Neural network algorithm to classify customer conversations based on sentiment classes (positive - for compliments, neutral, and negative - for complaints) on each SERVQUAL dimension. This method aims to obtain dimensions with the highest complaints as priority service improvements. This research uses conversation data of Paperlust subscribers on the “Chat With Us” service for the period January 1, 2020 to June 30, 2020. This study shows that 23% of chats received have positive sentiments, which are compliments and expressions of satisfaction with the services provided by Paperlust. 4% have negative sentiments, which are complaints and expressions of i disappointment with the services provided. The remaining 73% percent have neutral sentiments, which are common questions asked to customer service. The classification results on the SERVQUAL dimension show that Empathy, Responsiveness, dan Assurance dimensions receive more compliments than complaints. Meanwhile, the dimensions of Tangible and Reliability get more complaints than compliments, with the proportion of complaints above 90%. These results make the Tangible and Reliability dimensions an improvement priority for Paperlust.