The Challenge
A leading FMCG retail chain in Saudi Arabia operating across 7 warehouses and dozens of branches was losing revenue to chronic stockouts while simultaneously overstocking slow-moving items. With 5 years of fragmented sales data spanning thousands of SKUs, the team relied on spreadsheets and intuition for replenishment decisions. There was no central demand forecasting system, no S&OP process in place, and no visibility into branch-level demand patterns. Executive leadership recognized the need for a transformation: from gut-feel ordering to AI-powered, data-driven supply chain management.
Our Approach
- Conducted comprehensive data transformation integrating 5 years of monthly and weekly branch-level sales data across all SKUs
- Developed forecasting models using multiple ML approaches: ANN, XGBoost, Prophet, and polynomial regression for family and SKU-level predictions
- Built inventory optimization models with safety stock, reorder points, and service level targets per branch
- Created interactive Dash and Streamlit dashboards deployed to Heroku for real-time monitoring and decision support
- Implemented S&OP process with executive alignment meetings and structured demand review cycles
- Delivered commercial training program covering forecasting methodology, inventory control, and data-driven decision making
- Conducted geospatial analysis of branch network performance and demand distribution
Analytical Methods Used
Machine Learning Forecasting (ANN, XGBoost, Prophet)
Time Series Analysis & Regression
SKU & Family-Level Demand Disaggregation
Inventory Modeling (Safety Stock, ROP, EOQ)
S&OP Process Implementation
Geospatial Branch Analysis
Interactive Dashboard Development (Dash/Streamlit)
Tools Used
SCOPT AI
Python / Dash / Streamlit
Heroku Cloud Deployment
Key Outcomes & Results
- Delivered production-grade ML forecasting pipeline processing 5 years of multi-branch data
- Built interactive dashboards enabling real-time branch-level demand monitoring and inventory decisions
- Established S&OP process with structured executive review cycles
- Trained organization on forecasting methodology and inventory management best practices
- Created scalable forecasting framework with model comparison across ANN, XGBoost, Prophet, and regression approaches
- Deployed decision-support tools accessible to business users across the organization