Developing a framework for opportunity assessment of when to utilize machine learning to create data-driven products

Abstract: In!recent years, machine!learning has developed to the!extent that it can be utilized and implemented to create business value in organizations  by  either reducing costs or increasing  innovation and growth opportunities. Machine learning can unlock possibilities to create a better! product and experience, and thereby aid in gaining a stronger position in the industry. With millions of users traveling through their e2commerce platform, the case company of this thesis, a subscription based digital service company, has the potential tocreate an improved customer  experience using optimization and machine learning, generating business value and revenue. With limited resources and need for prioritization, understanding in which areas it would be most beneficial and generate most value to implement machine learning is critical. This thesis conducted an empirical study and thematic analysis based on semistructured interviews with machine learning engineers and managers at a subscription based digital service company to investigate how to assess when it is beneficial to utilize machine learning for optimization problems within an e2commerce organization. Impact, confidence, and effort were identified as suitable factors to assess the return on investment (ROI) of machine learning. In addition to this, three factors associated with machine learning were identified as required to have in place!or to consider in order to ensure a successful machine learning implementation. These three factors were data, business metrics (what to optimize), and discovery/research.