Using Text Mining Tools to Analyze Marketing Communication of Sales Representatives of a Distribution Company

Why do some sales representatives promote better and others promote worse? Are there linguistic predictors of sales success? Can computational linguistics methods help in identifying them? This thesis is devoted to the use of Text Mining tools to analyze the marketing communication of sales representatives of a distribution company. The relevance of the paper is due to the active digitalization of distribution companies and the increased demand for NLP tools in the marketing industry. Distribution companies work using the B2B (buyer-to-buyer) system: in order to promote products, the company's sales representatives need to communicate directly with customers, so they can directly influence the customer. Understanding the effectiveness and quality of the communication with customers is critical to optimize marketing efforts and improve results. Due to digitalization, large arrays of speech data recorded in marketing communication are gradually being formed. However, at the moment these records are not used in any way to improve the quality of service and increase sales, or are occasionally processed using expert manual analysis. In this context, the use of Text Mining tools becomes an important tool for analyzing marketing communication. The paper describes the experience of using computational and communicative linguistics methods on a real large-scale corpus of data provided by the company, identifies linguistic, communicative and paralinguistic signs typical for successful and unsuccessful sales representatives of a distribution company.