This study seeks to address the evident gap in empirical research concerning the sales generation mechanisms within Asset Management Firms, with particular attention to the influence of client phone interactions on the successful closure of opportunities. Utilizing predictive analytics—specifically feature selection techniques—the research develops a supervised learning model tailored to the distinct characteristics of sales processes in this sector. The primary aim is to identify key determinants within the sales pipeline that significantly affect forecasting accuracy, with a specific focus on the role of client communications. The study investigates a range of machine learning approaches, emphasizing the relevance of supervised learning for enhancing decision-making and commercial forecasting in asset management. The methodological framework is grounded in the CRISP-DM process, encompassing data preparation, cleaning, transformation, and modeling phases. Particular consideration is given to challenges such as handling missing and categorical data, as well as the critical role of feature selection and encoding techniques in improving forecast reliability. The proposed approach demonstrates the potential to enhance sales outcome prediction by quantifying the impact of client calls and isolating pivotal variables within the sales funnel.