01Open the case at the first negative point
Record case ID, item and alternate IDs, facility, exact location, inventory status, owner, lot or serial when applicable, base unit, negative quantity, first negative event and discovery time. Preserve screenshots or extracts only as supporting evidence and retain the transaction data source. Identify whether on-hand, available, allocated, projected or accounting inventory is negative; these are different measures. Determine which system and ledger own the balance. Stop uncontrolled adjustments, shipment or consumption for the affected scope when authorized without blocking unrelated good inventory. Capture users, interfaces and batch jobs active around the event. Define a narrow initial time window, then expand backward until a supported opening balance is found and forward through current state. Do not start from the current total alone.
02Rebuild the ordered transaction timeline
List opening balance and every receipt, return, transfer, putaway, status change, production completion, pick, pack, shipment, consumption, scrap, count, adjustment, cancel and reversal. Keep business event time, source creation, system posting and interface receipt timestamps separately. Sort each way to expose a receipt performed before shipment but posted later, a backdated adjustment, a replayed interface or a cancellation arriving out of sequence. Link original and reversal IDs and distinguish full from partial reversal. Check whether the same physical event posted in two systems or under two message IDs. GS1 EPCIS supports event time, record time, identified objects, location and business context; local ledger rules still determine quantity effects. Build a running balance by base unit and identify the exact transaction that crosses zero.
03Validate item, unit, location and status
Check item master, pack configuration, base unit, transaction unit, decimal precision, conversion factor, barcode mapping and effective date. UNECE Recommendation 20 provides coded units but does not define product-specific pack factors. Confirm source and destination locations exist, were active and allow the item and status. Review lot, serial, ownership, consignment, quarantine and available states. A warehouse may physically hold units that are unavailable or owned by another party; do not use them to cover a negative available balance. Test sign logic for returns, reversals and transfer directions. Look for one case posted as pieces and another as cartons, a stale pack factor, a child item consumed without assembly output or a transfer issue with no matching receipt. Preserve master-data history so current settings do not rewrite how old transactions were interpreted.
04Compare ledger with physical and document evidence
Count the defined item-location-status population under the cycle-count control and identify packages, lots or serials where required. Trace receipts to receiving records, shipments to package and vehicle events, transfers to both sites, returns to authorization and inspection, production to material issue and completion, and scrap to approval. UNECE INVRPT structures inventory quantities with movement details, location, date, status, owner and references; use these dimensions to avoid a facility-level net that hides location errors. Search adjacent locations, staging, quarantine, returns and goods in transit only through controlled records and safe physical checks. Do not move discovered stock before its current location and status are recorded. A physical match today does not prove the historical sequence; retain both count and transaction evidence.
05Classify cause and downstream impact
Classify timing delay, missing receipt, duplicate issue, premature shipment, unposted transfer receipt, wrong location, wrong status, wrong owner, unit conversion, sign error, interface replay, backdate, bad reversal, production posting gap, count adjustment error, unauthorized action or physical shortage. Record primary and contributing causes with evidence. Identify affected customer allocation, replenishment, purchasing, production, shipment, stock valuation, margin, availability promise, lot trace and supplier or carrier claim. Separate operational correction from financial or tax entries requiring qualified owners. Do not accuse a user of misconduct from one anomalous transaction. Check nearby items, locations and batch IDs for the same pattern because a configuration or interface defect may affect a population.
06Correct through controlled source records
Prefer correcting or completing the source receipt, transfer, reversal, unit or status transaction with authority and an audit link. If policy requires an inventory adjustment, record quantity, unit, location, status, reason, evidence, preparer, approver and effective time. Do not invent an offsetting receipt, borrow stock from another owner or change old timestamps to remove the negative. Prevent both the source correction and a temporary adjustment from remaining active. Reprocess downstream availability, reservations, replenishment, cost and order state as required. Test the corrected sequence from opening to closing balance and verify physical stock again when movement continued during review. Keep unresolved financial consequences with the appropriate owner rather than hiding them inside quantity.
07Close and prevent recurrence
Close when the running balance, source records and supported physical result agree, affected transactions are corrected once and downstream dependencies are reviewed. Record any residual difference and authority. Monitor negative-balance incidence, depth, duration, first-causing transaction, interface latency, unit errors, backdating, repeated adjustments and recurrence after correction. Add real-time or batch controls that reject or hold unsupported negative issues where appropriate, while allowing only documented business exceptions. Test the control against the confirmed failure and a valid late-posting scenario. Sample adjusted cases back to source evidence and sample negative ledger events for cases. Improve scanner flow, offline synchronization, transfer pairing, production completion, master-data change or reversal design according to root cause. Do not define success as zero visible negatives if users can hide them through uncontrolled positive adjustments.