Sales intelligence with data mining

Abstract

This technological validation project aims to meet one of the biggest challenges faced by companies that have a field sales team as a sales channel for their products: the lack of customer knowledge and market feedback to prepare a good commercial plan or quickly adjust the sales strategy during its execution. The field sales team, also known as the sales force or Sales Force Automation (SFA), is generally formed by a heterogeneous team, composed of its salespeople and commercial representatives with outsourced agents, which makes it difficult to collect information in the field and consequently the knowledge of the customer and the market. In the last three years, Orion has been researching and applying data mining, machine learning, and Artificial Intelligence (A.I.) techniques to social data from cities. During this period, experiments and proofs of concept were carried out by Orion's internal team to understand the validity of using these algorithms and techniques also in other company solutions, including the presentation of survey results to current customers, which proved their potential and commercial value. This research is defined by two main technological objectives. The first objective consists of expanding the pre-processing stage of data mining using three main groups of information: the partner company's sales database, the database referring to market research, and the database referring to sales targets. The second objective concerns the application of the following classes of algorithms on such bases: clustering (K-means) to define the segmentation of customers, having as reference their purchases and other data collected by the field team; association rules (Apriori) to define the untapped potential of a particular region, city or client, based on the socioeconomic and demographic characteristics of the location and the company's sales history; and finally, classification and outlier detection (KNN) to analyze the most and least sold products by region and customer segment, making it possible to determine, for example, which customers have not yet purchased from the collection and should have purchased. Based on Orion's experience with the implementation of commercial methodologies to increase the scale and volume of its commercial process, we conclude that the simple application of A.I., without a well-defined commercial strategy and monitoring methodology, will not achieve the potential gains of using the technology. Thus, Orion implemented in the last eighteen months a method called Sales Tank, created by the partner company Grandes Planos from tools and other methods already established in the market, such as ABM, SPIN, VOC and CTC Tree, and others. This method allows, from the increase in the level of knowledge of the customers in the dimensions: intelligence, engagement, competition, relationship, and qualification, to clearly determine the stage of the sale of each customer and to automate the actions that lead to the closing of a new sale, determining cadence and rhythm automates the commercial process. In summary, this technology validation project is structured on two main points: the first point is the application of a sales methodology that allows automation of the commercial process, guiding the field sales team, generating constant feedback to review the commercial planning and, mainly, guarantee that the commercial actions are effectively coordinated by the company itself. The second point is the application of artificial intelligence techniques in the various commercial stages: collecting customer and market data, planning a new product line, defining goals and analyzing sales made to automate actions, generating proactivity in the commercial process and enhancing results. (AU)