The research is focused on how companies manage and should manage implementation of artificial intelligence (AI) / machine learning (ML) based predictive analytics solutions. What are the preconditions of successful implementations and lessons learned from failed cases. Research methods used by the authors are literature review, experiments, qualitative and quantitative surveys. The aim of the research is to identify a model / framework to ensure successful implementation of AI solution in an organization. As a result of the study, the authors propose a model / structure that should be followed for the successful implementation of predictive analytics solutions. The main conclusion is that new IT initiatives, as predictive analytics, to some extent, is an IT project, must go through the classic stages of change management, just like any other initiative to change business processes. The authors recommend that companies allocate enough time, human and other necessary resources for the preparation and implementation of predictive analytics solutions. The goal should be clearly defined and measurable, and communication and change in business processes should be carefully planned. Another important step is training.