A bonded warehouse removes duty from the moment goods arrive, but it does not remove the goods from the customs record. That record is an electronic ledger, and it is maintained line by line, parcel by parcel, for as long as the stock sits inside the supervision zone. Brands tend to discover the ledger at the worst possible moment: during a write-off, when the quantity on the record no longer matches the quantity on the floor and somebody has to explain the difference.
This guide explains what the bonded ledger actually contains, how the three-document match works at the point of sale, where mismatches originate, and how a brand can keep the record balanced without turning every month into an audit. It is written for operators rather than for accountants, and it assumes the reader already sells into China through a bonded route.
Why the ledger sits on the brand’s risk register
The filing is normally done by the warehouse operator, because the operator holds the customs registration and the warehouse system. The exposure, however, belongs to the brand. The goods are the brand’s property, the duty liability attaches to those goods, and any gap discovered at write-off is resolved against the party that owns the inventory. A brand that treats the ledger as the operator’s paperwork is outsourcing a control while keeping the consequence.
The practical form of that consequence is rarely dramatic. It is more often a supplementary duty assessment, a delay while a record is corrected, or a temporary block on pushing new manifests for the affected product record. Each of those outcomes costs money at exactly the moment a brand is trying to sell.
What the electronic ledger records
A bonded ledger is a quantity account, not a financial account. It is organised by product record rather than by order, and every movement is booked against that record. Four movement types matter most.
Filing establishes the record. Each product record carries an identifier, a customs classification, a declared value, an origin, a specification set and a permitted unit of measure. Everything that follows is constrained by how that record was written. A record filed with a vague specification will generate questions at the first discrepancy, because there is no unambiguous basis for comparison.
Entry into the zone adds quantity. This is the inbound declaration, and it must agree with the physical receipt that the warehouse signs. The two documents are produced by different parties and reconciled later, which is why inbound receipt discipline matters more than it appears to.
Exit from the zone deducts quantity. In a retail bonded operation, exits are driven by individual parcels, so the ledger is decremented many times per day in small amounts. This is the highest-volume and highest-risk movement, because it depends on a continuous data feed rather than on a periodic document.
Adjustments close the gap. Returns re-entering the zone, transfers between records, destruction, re-export and approved corrections all move quantity without a sale. Adjustments require evidence, and the quality of that evidence is what a review will test.
The three-document match at the point of sale
Every retail parcel leaving a bonded zone is supported by a manifest, and that manifest is validated against three source documents: the order, the payment record and the logistics record. The match is not a formality. It is the mechanism that proves the sale was real, that the money moved, and that the goods physically travelled.
The three documents must agree on a small number of fields: the identity of the buyer, the amount paid, the item and quantity, and the shipment weight. When they agree, the manifest passes. When they disagree, the manifest is rejected or routed to manual review, and the parcel does not move until the disagreement is resolved.
Most brands lose control at the payment document. Payment is normally collected by the platform, settled on the platform’s schedule, and recorded under the platform’s own reference. The name on the payment record may be a masked account holder, the amount may reflect a platform subsidy or a coupon, and the settlement date may differ from the order date by days. None of this is wrong, but it means the payment record cannot be assumed to match the order without an explicit reconciliation rule.
Where mismatches actually come from
Discrepancies are rarely caused by theft or by bad faith. They come from ordinary commercial practice colliding with a quantity account that has no room for ambiguity.
The first cause is promotional pricing. A product record is filed with a declared value, and a campaign price may fall outside the tolerance band that the declaration supports. The order is genuine, the payment is genuine, and the manifest still fails, because the price no longer corresponds to the record.
The second cause is bundling. Three separately filed products sold as one gift set produce one order line and one payment, while the ledger expects three separate deductions. If the bundle is not mapped to its component records, the ledger either deducts the wrong product or deducts nothing.
The third cause is split shipment. One order dispatched as two parcels generates two logistics documents against one order and one payment. Unless the split is declared as a split, the quantity is counted twice or halved.
The fourth cause is cancellation after manifest push. The platform cancels, the manifest has already been submitted, and the ledger has already been decremented. The stock is physically back on the shelf but still recorded as sold.
The fifth cause is the unrecorded return. A customer return arrives at the warehouse, is put back into stock by the warehouse team, and never re-enters the ledger, because the return was processed as a warehouse event rather than as a customs event.
The sixth cause is dimensional disagreement. The logistics document carries a weight and a volume that do not match the product record, usually because packaging changed. The parcel is held, and the hold is recorded against the brand rather than against the packaging decision.
The seventh cause is identity inconsistency. The buyer name on the order, on the payment record and on the logistics document may be three variants of the same person. Masking rules differ between systems, and a strict comparison fails on formatting rather than on substance.
The eighth cause is the manual stock adjustment. A warehouse team corrects a picking error directly in its own system, the physical count becomes right, and the ledger silently becomes wrong. This is the most damaging cause, because it converts a small operational fix into an unexplained customs variance.
What the write-off cycle is actually testing
A write-off compares the ledger with reality over a defined period. The arithmetic is simple: opening quantity, plus quantity entered, minus quantity exited, plus or minus approved adjustments, should equal closing quantity. What the exercise tests is not the arithmetic but the classification of every item that does not reconcile.
A shortage can be a picking error, an unrecorded return, a manifest that deducted twice, or genuine loss. An excess can be an unrecorded inbound, a cancelled order that was never reversed, or a mis-mapped bundle. Each classification leads to a different remedy, and only some of them can be corrected by adjustment. A brand that cannot classify its variances will end up paying duty on stock it still owns, which is the worst of both outcomes.
The evidence standard matters as much as the number. A variance supported by a parcel-level trail, a signed count sheet and a documented correction is a routine matter. The same variance supported by an assertion is a finding.
The first seventy-two hours after a discrepancy appears
The instinct to correct the number immediately is usually wrong. The first action should be to freeze the affected product record so that no further manifests are pushed against it, because every additional parcel adds noise to an already unclear trail.
The second action is to pull the parcel-level trail for the period, not the summary. The summary will show a gap; only the parcel trail will show where the gap opened.
The third action is to classify and quantify before touching anything. A correction applied to an unclassified variance simply moves the problem into the next period, and a second unexplained movement is far harder to defend than the first.
The fourth action is to choose the remedy deliberately. Where the goods are physically present and the paperwork is wrong, an approved adjustment with evidence is correct. Where the goods are genuinely absent, accepting the duty is correct. What is never correct is pushing a synthetic order to balance the ledger, because that creates a second discrepancy in the sales record while leaving the first one unresolved.
Audit readiness: the document pack
A review is much easier when a fixed document pack exists before it is requested. The pack should contain the product record printouts, the inbound declarations, the manifest list for the period, the payment reconciliation, the inbound and outbound receipts, every adjustment approval with its supporting evidence, the count sheets, and the signed count minutes.
The pack should be assembled on a fixed monthly date rather than on demand. A pack built after a request is always incomplete, and an incomplete pack invites a wider review than the original question required.
A thirty-day plan to bring the ledger under control
Week one maps every product record to an internal item and a named owner. Records with no owner are the ones that drift, and the mapping exercise usually reveals more problems than the count does.
Week two instruments the three-document match. The objective is not to eliminate failures immediately but to log them, with a reason code, so that the distribution of causes becomes visible. Most brands find that two or three causes account for the majority of rejections, and those two or three are cheap to fix.
Week three runs a mock count and a mock write-off. The purpose is to discover what evidence is missing while there is still time to produce it.
Week four closes the gaps, writes the operating procedure, and fixes the monthly calendar for counting, reconciling and reviewing. The procedure should name who may approve an adjustment, because an approval right that sits with everyone is an approval right that sits with no one.
Questions worth asking a warehouse partner
Five questions separate a partner who runs a controlled ledger from one who runs a warehouse. Who files the ledger and who signs off adjustments? How quickly is a discrepancy notified, and in what format? What is the count cadence, and is the count independent of the picking team? Who bears the duty on a confirmed shortage? And can the partner produce a parcel-level trail on request, within a stated number of hours?
The answers matter more than the storage rate, because the ledger is the one place where physical stock and the customs record are forced to agree. A brand that treats it as a control rather than as paperwork protects its own margin, its own launch calendar, and its own ability to keep selling without interruption.