A reclass notice arrives on a Dallas to Charlotte shipment you tendered six weeks ago. The carrier re-measured the pallet at 48 by 40 by 54 inches, divided 450 pounds across 60 cubic feet, and rated the freight at 7.5 pounds per cubic foot instead of the 9.0 you quoted against. Your analyst opens the file to build a rebuttal and finds a bill of lading that reads one pallet, 450 pounds, with the dimension fields empty.
There is no argument to make here. Not a weak one, none at all. The carrier holds a measurement taken on its own equipment at its own dock on a dated shift, and you hold a recollection about freight that was delivered three weeks ago. A dispute is a contest between two records, and only one side brought one.
The carrier measures, and you assert
Nine inches of stack height is what separates those two density figures, and nine inches is roughly what a shrink wrap dome, a taller pallet than the spec assumed, or one carton turned the wrong way adds to a built unit. Your product master knows the carton. The carrier weighs and measures the handling unit that actually rolled off the trailer.
Old Dominion's account of data accuracy in LTL shipping puts numbers on the gap. Douglas Dynamics found that about 25 percent of its shipments had problems with length, weight, or classification, and learned it only after putting a transportation management system in. After the company trialled a dimensioner in Maine in the autumn of 2024, its weight and length based re-rates dropped to one percent. Bestorq's logistics manager in the same piece says as much as 25 percent of shipments were coming back with a re-weigh or re-classification, with up to 30 such invoices reaching him in a month.
Read the Douglas Dynamics figure carefully. The freight did not change when the dimensioner arrived, the record did. A quarter of shipments had been moving on numbers nobody had actually measured, and the carrier was measuring every one of them.
The 2025 scale raised the price of a number you may not capture
Classification used to absorb some of this. An item with a fixed class rated the same whether the pallet stood 45 inches or 54. That cushion is going away. Trucking Dive reported on 22 July 2025 that the NMFTA overhaul kicked off on 19 July 2025, moving freight with no handling, stowability, or liability concerns onto a single density based scale.
NMFTA's own published material on the change describes the Freight Classification Development Council's 11 sub density scale expanding to 13 sub-provisions, with sub 12 applying to densities from 35 but less than 50 pounds per cubic foot at class 55, and sub 13 applying to densities of 50 pounds per cubic foot or greater at class 50. The subject analyses published with Docket 2025-1 are blunter about where the obligation sits: density must be shown by the shipper on the shipping order and bill of lading at time of shipment.
That sentence is the one worth rereading. The scale did not ask carriers to work your density out. It asked you to state it, which means a blank dimension field is now a blank class justification.
Carriers gave shippers room to adjust. Trucking Dive also reported that FedEx Freight paused enforcement for 150 days, and that the carrier may apply an inspection surcharge under Items 980 and 981 where bill of lading information is incomplete or inaccurate. Under the rules tariff cited there, a reweigh differing by less than 50 pounds carries no fee, and a larger difference draws 38 dollars. Note what the tolerance attaches to. It forgives a weight you declared and got slightly wrong, and says nothing about a dimension you never declared, because there is no figure of yours to compare against.
Reconstruction is not evidence
The instinct when the reclass lands is to go and measure one. Somebody walks to the warehouse, builds a unit of the same SKU, finds 45 inches, and photographs the tape. That is a demonstration of what the pallet can be. It is not a record of what that pallet was.
A carrier pricing desk will say so in one line, and it will be right. Different day, different packer, different wrap tension, possibly a different pallet supplier. Your photograph is dated today and the shipment moved in September. Nothing produced after the freight left the dock is contemporaneous, and that is the only quality that matters once two parties disagree about a physical fact.
So the outcome is set at tender, not at invoice. Every control your audit function owns sits downstream of the moment the evidence had to be created. An audit can confirm the carrier did its own arithmetic correctly. It cannot put a tape measure on a pallet that was unloaded last month.
What a defensible tender record holds
The bar is lower than most teams assume. A record that survives a carrier challenge contains:
- Measured dimensions of the built handling unit, with the pallet, any overhang, and the wrap included, rather than carton dimensions copied from the product master.
- The weight from the same event, with its timestamp, so density is computed from two figures taken minutes apart.
- What took the measurement and when: a dimensioner, a floor scale with an ID, or a named operator and a tape.
- The NMFC item and sub-provision you rated against, plus the density you asserted, stored as a number rather than inferred later from a class code.
- A photograph of the finished unit before it is loaded.
Nothing on that list is exotic, and in most operations four of the five already exist somewhere: a dock app, a scale printout, a WMS field that never maps into the tender payload. The gap is rarely measurement. It is that the measurement never reaches the bill of lading.
What Penny does with the reclass line
Penny audits the invoice before it is paid, and on a classification correction she works three questions in order.
First, does the carrier's billed class follow from the carrier's own figures? She recomputes density from the cited dimensions and the re-measured weight and checks whether the result actually lands in the sub-provision that was billed. Rounding at a band edge, a cubic foot conversion applied twice, or a dimension transposed between the inspection note and the invoice are all recoverable, and they are recoverable on the carrier's evidence rather than on yours.
Second, does your tender record contain a measured dimension at all? Where it does, there is a real dispute with two numbers in it and Penny drafts it. Where it does not, she does not invent a position. The line comes back marked needs-data with the missing field named, which is the behaviour described in an agent that says needs data. An agent that fabricates a 45 inch height to argue with costs you more than one that tells you plainly that the argument does not exist.
Third, has this SKU done it before? A part number reclassified on eleven of twelve moves is not twelve disputes. It is one wrong number in the product master, and it belongs to the same pattern logic set out in the same 400 dollars every month. Charges that keep arriving on schedule have a cause with an owner, and in this case the owner is whoever maintains item dimensions.
What to do next
Pull 90 days of LTL invoices and isolate every line carrying a class or weight correction. For each one, ask a single question: did your tender record contain a measured dimension for that handling unit. Count the share where the answer is no. That share is not disputed spend or lost spend, it is undefendable spend, and it is almost certainly the largest of the three.
Then take the lines where you did have a dimension and recompute density from the carrier's cited measurements. The recovery rate on that second group tends to surprise people, because carriers make ordinary arithmetic mistakes and almost nobody checks their working.
Bring those 90 days to a walkthrough and we will run Penny across them live, sorting each reclass into recoverable, defensible, and owed, with the missing field named on every line in the third bucket.