To do well in sales we need to understand what drives them. This is important for making marketing plans and keeping sales up over time. We are trying to figure out how much each marketing method contributes to sales when customers take a path to buy something. We also want to make a system to predict future sales. We suggest using a combination of Multi Touch Attribution and advanced Machine Learning to make predictions our method involves looking at three ways to give credit to marketing methods. Linear, Time Decay and Position Based. To break down past sales data into what each channel did. We then use the method to help make a customized system to predict sales using XGBoost and a Neural Network to make the predictions more accurate. We tested this using data from an e commerce site with 50,000 customer journeys across eight marketing channels the results show that using the Position Based method with our suggested system works best giving us an idea of how well we can predict sales. This helps us understand what marketing methods work and makes it easier to predict sales so we can make plans. This study shows that combining attribution science with intelligence is a good way to make a system that works well for businesses, with many sales channels. Sales Attribution Modeling, Predictive Sales Analytics, Multi Touch Attribution, Ensemble Learning, XGBoost Marketing Mix Optimization, Machine Learning, Customer Journey Analytics, Revenue Forecasting. 100-114 Smart Innovations in Computer Science and Applications Special Issue Sakshi Raut