Supply Chain & Inventory Analytics Case Study 01
Winona State University

Rebuilding reorder points around real demand variability

Redesigned reorder-point logic across three SKU categories using a demand variability analysis — cutting excess inventory by 18% without moving the stockout rate.

Project Lead Winona State University
Scope 3 SKU categories
Method CoV-based demand clustering
Validation 3-month order backtest
18%
Reduction in excess inventory across the three SKU categories
$42k
Estimated working capital freed from carrying-cost savings
1.8%
Stockout rate held — essentially flat versus baseline
§ 01Challenge

One rule, applied to every SKU

Warehouse carrying costs were climbing because reorder points were set from average demand rather than true demand variability. The same static rule governed every SKU — overstocking slow movers while leaving fast movers exposed.

The result was a system that looked balanced on paper but was quietly wrong in both directions at once.

§ 02Solution

Segment by volatility, not by category

Pulled 12 months of sales and lead-time data, then clustered SKUs by coefficient of variation (CoV) in demand. Each volatility segment got its own reorder point and safety stock calculation, replacing the single blanket rule.

The new model was tested against three months of live order data before rollout, confirming it held up against real ordering patterns rather than just historical averages.

§ 03Outcome

Leaner shelves, same service level

Excess inventory across the three SKU categories dropped 18%, translating to an estimated $42,000 in freed-up working capital. Stockout rate stayed at 1.8% — essentially flat from baseline — confirming the leaner inventory didn't come at the cost of service.

Reorder point, before vs. after segmentation

Illustrative inventory position by SKU volatility cluster
Low CoV Medium CoV High CoV old demand (noisy)
Actual demand Single static reorder point Reorder point per volatility cluster