AI-driven supply chain replenishment: safety stock, demand forecasting, transfers and procurement
7 articles
For seasonal and promotional goods the hard part is rhythm, not formulas: build-up, peak and clearance phases each have different objective functions. Seasonal indices, uplift estimation, phased replenishment and clearance tactics.
A Kingdee Cloud walkthrough of the full loop: suggestion → plan confirmation → order splitting → purchase order Save/Submit/Audit → arrival backfill, with field mapping, idempotency and retry design.
Stockouts and overstock are two faces of the same coin. Five recurring pitfalls in replenishment models — uniform parameters, the average trap, promotion contamination, one-sided KPIs and stale parameters — and how to correct each.
When total stock suffices but one warehouse stockouts, transfer before purchasing. Decision order, rebalancing triggers, safety-stock allocation across warehouses, and the data calibers that make transfers executable.
A practical comparison of moving average, exponential smoothing (SES/Holt/Holt-Winters), Croston and machine-learning methods for e-commerce replenishment, with SKU-tiered selection guidance.
The statistics behind safety stock: the classic formula, z-scores per service level, the combined demand-and-lead-time variability formula, plus a worked example and code.
Why experience-based ordering breaks down, what data-driven replenishment actually consists of, and a pragmatic maturity path for e-commerce and retail teams.
