Design and Deployment of ML in CRM to Identify Leads

Abstract

In today’s era, organizations are increasingly prioritizing process automation to optimize efficiency and drive sales. One area where Machine Learning (ML) techniques can be particularly valuable is in automating tasks such as lead classification for sales. In Jupyter Notebooks, logistic regression was utilized to design and to train a model to accurately predict whether a lead will convert into a client or not. Then, in Azure Machine Learning Studio which is a Machine Learning Operations platform (MLOps), the Two-Class Logistic Regression algorithm was used to design a pipeline, train a model, and deploy a web service, which is consumed by Salesforce system through an Apex code. The web service receives the variables of a particular lead record and then returns the prediction as a numeric ranking. By leveraging these ML techniques, firm’s resources can strategically be focused for maximum effectiveness. Overall, our work involves a C# windows application to extract CRM marketing interactions, leveraging the power of ML, a logistic regression model in AML and Apex code. This approach enables us to drive efficiency, enhance sales outcomes, and allocate resources more effectively.