From Data Ingestion to Decision Support: An End-to-End Pipeline for Modern Sales Intelligence Systems

This paper presents SalesVision 360, a complete, production-grade data engineering and machine learning pipeline designed to unify sales analytics, demand forecasting, and customer segmentation within a single operational platform. The system integrates automated data scraping, synthetic large-scale sales generation, Snowflake data warehousing, dbt-based data modeling, time-series forecasting models, behavioral segmentation using K-Means, and a dual-dashboard environment (Power BI and Streamlit). Using real-world structured patterns, the pipeline achieves highly accurate predictions for daily sales revenue (MAPE ≈ 3.7%) and identifies five customer segments with strong business interpretability. The objective is to demonstrate a modern, scalable, enterprise-ready analytics ecosystem suitable for retail, e-commerce, and omnichannel businesses. The methodology, system architecture, model performance, and implications for data-driven organizations are discussed.