The distributor faced high overhead costs due to excess warehouse inventory, alongside frequent stock-outs of high-demand items that led to lost sales. Purchasing decisions were made on static reports and gut feelings. Additionally, the company had no objective way to track supplier performance, lead times, or shipment fulfillment rates, making it difficult to negotiate vendor terms or prevent supply bottlenecks. Our strategy was to replace gut-feel purchasing with data-driven decisions — by giving the operations team a system that tells them exactly what to order, how much, when, and from which supplier.
The root cause of both over-stocking and stock-outs was the same: the business had no accurate forward-looking view of demand. They were reacting to what already happened instead of preparing for what was coming. We needed to introduce predictive intelligence into the supply chain and pair it with objective vendor accountability.
We adopted a three-part strategy: Predict → Optimize → Score.
→ Predict Demand: Build a time-series forecasting engine that accounts for seasonality, promotions, and market trends — so purchasing decisions are made on future demand, not past averages.
→ Optimize Stock Levels: Use the forecast to automatically calculate the exact right safety stock and reorder points for every product — eliminating both over-ordering and stock-outs simultaneously.
→ Score Suppliers: Create an objective vendor scorecard so the business knows which suppliers are reliable and which are creating bottlenecks — enabling smarter negotiations and contingency planning.
What we built to execute this strategy:
1. A Python-based demand forecasting engine using Facebook Prophet, trained on 18 months of sales history and tuned for seasonal spikes and promotional events.
2. Dynamic reorder point calculations that update automatically as demand forecasts shift — no more manual reorder level reviews.
3. A Power BI Vendor Scorecard tracking on-time delivery rates, average lead times, damage/return rates, and cost variance for every supplier.
4. An Inventory Health Dashboard surfacing slow-moving stock, overstock alerts, and stockout risk scores by SKU.
5. Daily AI-generated purchase order recommendations — the system tells procurement exactly what to buy, in what quantity, and from which vendor.
The outcome: the team stopped debating what to order and started confidently executing smart purchase plans — backed by 92% accurate forecasts.