Businesses are required to constantly improve their products and services if they need to stay in the market. One way to do the same, is collecting customer feedback about their products and services. Our client was in the business of helping their customers to find the best marketing strategy based on the sentiment of their customer’s feedback. The feedback collection was large and to manually read, address and extract useful information for improvement proved to be a humongous task. There was a requirement for a quicker and efficient way to assess the feedback sentiments and address problems that were necessary. They wanted us to build a system to estimate the customer sentiment about their services in order to develop a better customer servicing plan and a better marketing strategy for their service. This required study and classification of the comments to categorise them and segregate them to extract useful information.
The objective of the project was to identify positive, negative and neutral sentiments from the feedback comments. We developed a Natural Language Processing model to classify a feedback comment into these three buckets. We then developed a reporting module to automatically trigger alerts on the clients inbox when they received negative feedback about their service.