Every landed cost figure is built from supplier invoices, and supplier invoices contain wrong quantities, stale prices, and charges that belong to other orders. This guide shows how data errors travel into landed cost and how to catch them at the gate.
Price variance is the commonest: the invoice price differs from the agreed price by a small percentage that nobody queries because the total looks plausible. Quantity errors follow, shipped short or over with the invoice following the order rather than the packing list. Then come allocation errors, where a charge covering three shipments is loaded onto one, making one product look ruinous and two look miraculously cheap.
Currency and unit errors are the quiet killers. A price agreed per carton invoiced per piece, or a total converted at a rate from the wrong month, distorts unit cost by factors rather than percents. These are also the easiest errors to catch automatically, because they produce numbers wildly unlike the product's history.
The classic control still works: match the purchase order, the receiving record, and the invoice before payment and before costing. Price against the order, quantity against what actually arrived, and charges against the shipment they belong to. Discrepancies pause the line, not the whole invoice, so one bad line does not delay payment for fifty good ones.
For landed cost specifically, add a fourth check at entry: the customs entry's values and quantities against the same records. The entry is what duty was charged on, and a mismatch between the entry and the commercial records is both a costing error and a compliance question in waiting.
Nobody can eyeball every line, so build the exceptions: unit cost moved more than a set percentage against the trailing average, quantity variance against receipts, unallocated charges older than a threshold, credit notes unmatched after thirty days. Reviewing thirty exceptions a week protects the data better than sampling three hundred clean lines.
Close the loop with suppliers. Recurring errors from one factory, always in the supplier's favour, are a conversation backed by a variance report. Most suppliers fix their invoicing quickly when the buyer demonstrates, politely and with data, that every invoice is being matched.
Pricing, reorder, and supplier decisions all read landed cost. A phantom cost increase can kill a healthy product; a hidden one can keep a losing product on life support for seasons. When the data quality work lands, the first visible effect is usually a handful of products whose true margins surprise everyone, in both directions.
Treat data quality as a standing control with an owner and a monthly variance summary, not a cleanup project. Invoice errors are generated continuously by ordinary business, so the checking has to be continuous too.
Keep reading: How can landed cost data improve your next Incoterms negotiation with suppliers?, How do you benchmark your landed cost against the market without competitor data?, and How do finance teams build an audit trail for landed cost? cover the nearby cases in this series.
Checking the invoice against the purchase order and the receiving record before payment: price against what was agreed, quantity against what actually arrived. For importers, matching the customs entry against the same records adds a fourth check that protects both costing and compliance.
Allocation errors and unit or currency errors. Loading one shipment's charges onto another product, or mixing per carton and per piece pricing, distorts unit costs badly and invisibly, while small price variances mostly add noise.
They focus human review where errors actually live: unit costs that jumped against history, quantities that disagree with receipts, stale unallocated charges, and unmatched credit notes. Reviewing exceptions catches more errors than sampling clean invoices.
Continuously at the matching gate, with a monthly variance summary for patterns by supplier and product. Errors are created by everyday business, so one off cleanups decay within a quarter.