By Jacob Hopewell, nShift

Ask a finance team how they control supplier spend and you will hear a version of the same answer.

An order is raised, goods are received, the invoice is matched against both, and anything that does not line up gets held. It is unglamorous and it works.

Freight is the exception. There is no purchase order for a parcel. The shipment is booked, it moves, and the invoice arrives weeks later carrying rates, weights, and a stack of surcharges that were never agreed line by line in advance.

The reference point that protects every other category of spend does not exist, so the invoice is approved on trust and paid.

That gap has a price. Published market research puts recoverable overcharges at 3 to 7% of total freight spend, with averages between 3 and 5% and the highest documented recoveries reaching 7%.

DataIntelo and MarketIntelo report the range across retail, manufacturing, and healthcare. Forrester quotes an IT director at a manufacturer with billions in annual freight spend saying audit initiatives typically capture 1 to 6% of savings when every invoice is reviewed.

For a retailer spending 10 million a year moving goods, the middle of that range is half a million pounds sitting with carriers.

Errors are the normal state, not the exception

The scale is easy to underestimate because freight billing errors are rarely dramatic. They are small, frequent, and individually defensible.

The Journal of Commerce freight benchmark study, a survey of 77 shippers, found an average invoice inaccuracy rate of 20.8%.

The sample skews enterprise and the study is from 2020, so treat it as a signal rather than a forecast for your own operation. The wider evidence points the same way.

Ardent Partners put the average accounts payable invoice exception rate at 18.4% in 2025 across all invoice types. Drewry, looking at container shipping, found error rates of 2 to 5% for larger shippers on annual contracts, rising to 30% for smaller shippers buying on the spot market.

Disciplined contracts still leak, and ad hoc buying leaks badly.

A single shipment can carry five to ten surcharges, each calculated differently by carrier and by country.

Chargeable weight can be volumetric or actual, and the two disagree constantly. Multiply that by hundreds of thousands of lines a year and manual checking covers a rounding error of the total.

The economics of not checking

Here is why most teams stop. Investigating one questionable line costs more than the line is worth.

Benchmark figures put in-house freight invoice processing at around 18 dollars per invoice, against roughly 6 dollars when the work is outsourced.

Add the time to pull the contract, find the shipment record, work out the correct chargeable weight, and write to the carrier, and a 4-pound discrepancy is not worth a person’s afternoon.

Across 11 interviews we ran at nShift with shippers this year before launching nShift Audit, one line came up again and again: it is easier to absorb it than fight it.

So, the leak is a rational response to a checking cost that exceeds the value of any individual error. The problem only becomes solvable when that cost approaches zero.

The reference data already exists

Most shippers already hold the missing half of the comparison. If you book and print through a shipping platform, you have the shipment record: what was collected, what it weighed, what service was bought, whether it arrived on time.

Set that beside the contracted tariff and the carrier invoice, and you have a three-way match that behaves the way purchase order matching does in every other category.

That third leg matters more than it sounds. Comparing an invoice against a rate card tells you whether the price was right.

It cannot tell you that a shipment was cancelled and billed anyway, that a duplicate went out, or that a service failure was charged at the premium rate. Those need evidence of what actually moved.

It is the principle we built nShift Audit on, which becomes generally available in October.

It reads carrier invoices in their raw formats without per-carrier setup, checks every line against the contracted tariff and the shipment record already held in the platform, and flags each deviation with its monetary impact and a plain explanation of what went wrong.

The design point holds whether or not you use it: an audit is only as good as the shipment truth you can put beside the invoice.