Start with these five checks.
- Reconcile on-hand, reserved and available-to-promise stock.
- Measure stockouts by SKU, location and day.
- Separate forecast, supplier, allocation and execution causes.
- Segment priority items, launches and long-lead products.
- Check whether inventory records match physical counts.
Ask for only what can change the answer.
- Inventory eventsReceipts, picks, reservations and adjustments
- DemandOrders, lost sales and substitutions
- SupplyPurchase orders and actual lead times
- NetworkLocation, capacity and transfer time
- EconomicsMargin, holding cost and markdown
- PolicyService target, reorder and allocation rules
Test one reversible move.
Freeze a historical period and reproduce its actual inventory path. Replay one policy for a priority SKU class using the same daily information and constraints. Compare service level, working capital, expedites and aged stock; reject the rule if it shifts stockouts to priority locations or consumes unacceptable cash.
A decision your team can use.
- 01A reconciled stock ledger
- 02A stockout cause tree
- 03SKU-location opportunity sizing
- 04A time-correct policy replay
Common questions.
Why do totals hide the problem?
Aggregate inventory can rise while priority SKUs or locations remain unavailable.
What grain should be used?
SKU-location-day is a practical default; faster networks may need event-level timing.
Should safety stock be raised first?
Not before placement, allocation, lead time and record accuracy are checked.