This feature is like a smart crystal ball built specifically for items that sell infrequently and unpredictably (spare parts, slow movers). Instead of pretending they sell every week, it predicts when the next order is likely to happen and how big it will be, so planners can stock just enough without overfilling the warehouse.
Traditional forecasting methods perform poorly on intermittent or ‘lumpy’ demand (e.g., spare parts, slow-moving SKUs), causing either chronic stockouts or expensive overstock. Croston’s method in Dynamics 365 provides a better forecast for these items so supply chain and retail operations can plan inventory and replenishment more accurately.
Embedded into Dynamics 365 Supply Chain workflows and data model (item-level demand history, master planning), making it sticky for existing Microsoft ERP customers and hard to displace once configured.
Classical-ML (Scikit/XGBoost)
Time-Series DB
Medium (Integration logic)
Forecast quality depends on the volume and cleanliness of historical demand data and the appropriateness of Croston’s assumptions for each SKU; poor data or misclassification of item demand patterns will limit effectiveness.
Early Majority
Compared with generic time-series forecasting in ERPs, this focuses specifically on intermittent demand (classic Croston’s method) and is natively integrated into Dynamics 365 Supply Chain Management, giving planners a specialized, out-of-the-box option for spare parts and slow-moving SKU forecasting in retail and manufacturing contexts.