Summary
Wrong weights and dimensions cost you money in six places: storage, picking labour, freight, transport, dock time and stalled projects.
Every warehouse holds these numbers somewhere. Almost none of them were checked last week, because nothing in the process ever asks whether a dimension is still true. Suppliers change packaging and nothing leaves a trace.
The bill is real and it is large. It just never shows up as a line anyone owns, because item data is never used where it is created. Six teams read the same field, each one feels a different symptom, and none of them owns it.
Here is where the money goes.
Storage and labour: two costs that never reach an invoice
Two of the six buckets sit entirely inside your four walls. Neither ever lands on an invoice, which is why they can run for years without anyone noticing.
Worth knowing how common the underlying problem is. Research published by ECR Retail Loss in December 2025 found that over 60% of inventory records contain inaccuracies, and that correcting them produced sales uplifts of 4% to 11% in a field experiment.
That study looked at quantity on hand, not dimensions, so it says nothing directly about your cube data. What it does show is that inaccurate records are the normal state of a warehouse database rather than the exception, and that cleaning them up pays.
Storage: you pay rent on air
Slotting and capacity planning both run on volume. Overstate an item and the system reserves space you never use. Understate it and you get overflow, honeycombing and rehandling.
Careful with the blame here, because racking fill has several parents. Honeycombing, seasonality, min and max rules and slotting policy all leave gaps of their own. What bad dimensions add is that you can no longer tell which one you are looking at. A policy problem and a data problem look identical when the numbers underneath are guesses.
Check it: pull your slot occupancy report and compare reserved cube against measured cube for your top 200 movers. The gap is rent on air.
Labour: every wrong dimension gets fixed by hand, later
Hardly anyone traces this one back to the item master, but it starts there.
Pick to box only works if the system knows what fits in which box. Without dimensions it cannot choose, so either the picker decides at the face or everything goes into the biggest option and gets sorted out at the bench. Pick to pallet runs into the same wall with weight and stackability. Building a pallet as you pick means knowing what can go under what. Without that, the pallet gets built twice.
Clustering costs the most and shows the least. A WMS batches orders onto a cart or into a cage using the volume it thinks those orders take up. If the volumes are wrong, the cart either leaves half empty or fills up before the route is done and the picker walks it again. The route was never as good as the report says it was.
The compounding version is easy to spot at the loading bay. Three cages come off the pick face for one store, someone eyeballs them and squeezes them into two. That is a person being paid to correct an item master field.
Check it: stand at the pack bench for an hour and count re-packs. Then ask the shift lead how often cages get consolidated at the bay.
Freight: wrong box, then a reweigh you cannot argue with
This bucket has the hardest evidence behind it, because carriers publish what they do.
Cartonization is only as good as what you feed it. Give it estimates and it will recommend boxes that do not close, or boxes with a third of their volume in air, and packers stop trusting it inside a week. After that they pick boxes by eye, and you are paying for software nobody uses.
Box choice is not only about fit. It decides whether you stay under a carrier's size threshold, which rate band you land in, and whether an oversize surcharge kicks in. None of that arithmetic works on a guessed dimension.
Then the parcel reaches the carrier's hub, passes a certified dimensioner and a scale, and gets reweighed. Where their number and yours disagree, theirs wins. You get a dimensional weight adjustment, usually with a correction fee attached, on an invoice that arrives weeks later when nobody remembers the shipment. We have written before about what a wrong measurement does to a freight invoice.
Those figures come from a management alert published by the USPS Office of Inspector General on 5 May 2026, report number 25-130-1-R26. They are easy to read the wrong way round. They do not say most parcels are mismeasured. They say the largest parcel carrier in the United States cannot check most of what it carries, so whatever the shipper declares is usually what gets billed. That works in your favour right up until it does not.
Errors run both ways, and that is the case nobody plans for. When the measuring equipment itself failed for three months in 2026, a follow-up OIG report from 29 July 2026 put the damage at roughly $22.6 million in undetected underpaid postage and $6.1 million in overpaid postage. If your own numbers are guesses, you have nothing to audit an invoice against, so you will never know which side of that split you are sitting on.
Dimensional weight: the divisor moved in July 2026
Carriers charge for space as well as mass. Dimensional weight is length times width times height, divided by the carrier's dimensional factor, and the carrier bills whichever is greater: actual weight or dimensional weight.
That factor is not universal. It changes by carrier, by rate table, by region and by contract:
| Carrier or mode | Divisor | Scope |
|---|---|---|
| UPS, US Daily Rates | 139 in3/lb | 166 for Retail Rates. Every package, no size threshold |
| FedEx, US domestic | 139 in3/lb | Every package, no size threshold |
| USPS | 139 in3/lb | Was 166 until 12 July 2026. Only above 1 cubic foot. Fractions now round up |
| DHL Express and UPS in Europe | 5,000 cm3/kg | Roughly equivalent to 139 in3/lb |
| Air freight (IATA) | 6,000 cm3/kg | Whichever is higher, volumetric or actual |
| European road groupage | No divisor | Conventions instead: commonly 333 kg per m3, or 1,850 kg per loading metre |
View the carrier divisor table on desktop for the full data.
The USPS change is worth getting right, because a lot of the coverage got the scope wrong. The divisor moved from 166 to 139 on 12 July 2026, and a second change came with it: fractional measurements now round up to the next whole inch instead of to the nearest one.
But USPS still applies dimensional weight only to parcels over one cubic foot, while FedEx and UPS apply it to everything. So the number now matches what the big two use, but it reaches a smaller share of parcels. The current rules sit on the UPS shipping dimensions and weight page, which also shows the split between the 139 Daily Rate divisor and the 166 Retail Rate divisor.
For European operations, 139 in3/lb works out at roughly 5,166 cm3/kg. Above its threshold, USPS has effectively moved into line with the European parcel norm of 5,000. The direction is the same everywhere: space costs more, and the tolerance for a wrong dimension keeps shrinking.
Oversize surcharges: one centimetre, six times the fee
Dimensional weight is the gradual part of the bill. Surcharges are the cliff, and this is where a small data error stops being proportionate to its size.
Every parcel carrier sets thresholds. Cross one and a flat fee lands on the parcel, no matter how far over you went. Girth makes it worse: length plus girth counts width and height twice, so a one centimetre error on a side that is not the longest moves the total by two.
Three worked examples, using published 2026 thresholds:
| Threshold | Recorded | Actual | Cost of the difference |
|---|---|---|---|
| DHL Express Europe, oversize above 100cm longest side | 100 x 60 x 40 cm, no surcharge | 101 x 60 x 40 cm | EUR 20.00 per piece, every time |
| UPS Germany, large package above 300cm length plus girth | 150 x 40 x 35 cm, EUR 23.65 | 150 x 40 x 35.5 cm | EUR 101.80, plus 40kg minimum billable |
| UPS US, large package above 130in length plus girth | 42 x 22 x 22 in, $38.50 | 42 x 22 x 22.5 in | $273.00, plus 90lb minimum billable |
View the surcharge threshold table on desktop for the full data.
The shape of the curve matters more than the exact numbers, since your contract rates will differ. Cost near a threshold is flat and then vertical, not proportional to size. In the UPS example, a 0.4% error in one dimension multiplies the accessorial charge on that parcel roughly six times, and it repeats on every parcel of that SKU until somebody re-measures the carton.
Thresholds are published, so they are worth checking against your own range. DHL Express lists its oversize and non-conveyable triggers, and UPS and FedEx publish theirs. DPD and GLS have no single European tariff, so their thresholds vary by country and sit in your own rate card.
For the item master this means something narrow and specific. Any SKU whose carton sits within a few centimetres of a threshold needs a measured value rather than a supplied or estimated one, and it needs re-checking whenever the packaging changes. That is usually a short list, and it pays back faster than anything else in the file.
Check it: ask finance for one month of carrier invoices and filter on adjustment and correction lines. Sort by SKU. The repeat offenders are your list.
Transport and dock: trailers leave with air, stock waits to go live
Transport: you pay linehaul on the planner's safety margin
Two things decide what goes on a trailer: the floor space each unit takes up, and the weight it puts on each axle. Volume on its own answers neither.
A unit with a perfectly ordinary cube can still eat two pallet positions on the trailer floor, either because it is three metres long or because the orientation arrow on the side means it can never be stood on end. Cube tells you what fits in a box. Floor space is a different question, and weight distribution is a legal one rather than an optimisation one. Get that wrong and it becomes a roadside problem instead of a warehouse problem.
Without reliable weights you also cannot say whether a lane cubes out or weighs out, which is the first question in any transport review. So planners add a safety margin. That margin is empty space, on every trailer, every day, and you pay the full linehaul rate to move it.
Trailer cost is close to fixed, which makes the arithmetic unforgiving. A trailer at 80% fill costs almost exactly what the same trailer costs at 100%. Every point of fill you give away to a safety margin comes back as extra movements, and those movements are the biggest single line in most outbound transport budgets.
In Europe there is now a deadline attached. Under the EU packaging regulation, the empty space ratio for grouped, transport and e-commerce packaging is capped at 50%, applying from 2030 at the earliest depending on when the implementing acts land. At that point empty space stops being only a cost and becomes a compliance line. You cannot report a ratio you cannot calculate, and you cannot calculate it from estimated dimensions.
Check it: take last month's outbound loads and look at average fill against planned fill. The difference is the safety margin.
Dock: put-away waits for a tape measure
New lines arrive with no usable data. Put-away needs a destination, the WMS cannot pick one until it knows the size, so somebody walks over and measures it while the pallet sits.
At low volumes that is an annoyance. On a contract taking in thousands of new lines a season it turns into detention, delayed put-away and stock that is on site but not sellable.
A faster measuring process at the desk does not fix it. Measuring where the item is already being handled does, and it takes seconds instead of a trip. You can see what that looks like at an inbound station.
Check it: measure the hours between goods receipt and stock going sellable on new lines, not on repeats.
Projects stall before they start
The business case waits for the data
Every automation, slotting or relayout project starts with the same two questions: what is going in, and how will it be handled. Teams with an item master they trust can answer both and get started.
Teams without one spend the first phase building the data under time pressure, usually with consultants on the clock, which is the most expensive way to do this work. A warehouse profiling exercise has the same problem. Profiling reads the item master, so it inherits everything wrong with it.
Check it: look at the last automation or slotting project and count the days billed before anything was designed.
And some values are just wrong
Every item master holds a handful of numbers that are visibly impossible. An ordinary carton listed at a tonne. A pallet that would not fit through the door.
They survive because nothing in the data tells you which values to distrust. A dimension captured with a laser and a dimension typed at speed by somebody covering a shift look exactly the same once they are sitting in a database. With no obvious first item on the list, nobody starts.
Why it compounds: six teams, one field
Each of these lands on a different team. Storage blames the layout, picking blames the routes, packing blames the box range, transport blames the carrier. Nobody looks at the field they all read from, because from where each of them sits it looks like their own problem.
That is also why it rarely gets funded. The cost is real and large, and it is spread so thinly across so many budgets that it never turns into a line anyone owns.
The other side of the transaction shows what it adds up to. Since automated package verification started in 2017, it has increased USPS postage collection by $1.1 billion. That is a carrier's gain, and it came out of shippers' pockets one small adjustment at a time, with no single adjustment big enough for any one shipper to chase. The money is not hypothetical. It just leaves in pieces too small to notice.
So why has nobody fixed it?
Not for lack of knowing. There are three ways to get this data, and each one has a catch.
Suppliers send it
Cheapest, and the least reliable. Coverage is patchy, packaging changes never get communicated, and the figures often describe the product rather than the shipping unit. Distributors either work with what they were given or capture it themselves.
A fixed dimensioning station
The proper answer, and rarely bought, because of everything that has to happen first. The vendor is not in your ERP, so procurement starts cold. Someone specifies it, someone installs it, someone owns it, and the warehouse does not want to own capital equipment that is not moving product. Someone gets trained and rostered onto it. On the quotes customers show us, it runs into tens of thousands. Every step is reasonable, and together they are a wall. The project is never rejected. It just never starts.
There is a physical limit too. A fixed station only measures what passes the spot where it stands, so everything already sitting in your racking never gets measured at all.
Check the specification before you assume it covers your range. From the evaluations we sit in on, these machines have a minimum measurable size as well as a maximum, usually a few centimetres, so the smallest items in your catalogue may fall outside it. Many do offer irregular or polybag modes, which is worth knowing before anyone claims they cannot handle soft goods.
A tape measure and a scale
Where most operations actually are. It works, and it carries an error nobody quantifies. Two people measure the same carton differently. Nobody measures the same way twice on a Friday. Soft goods change size depending on how hard you press.
A tape also gives you a number with no record of who took it, when, or how, so you cannot check it later even if you want to. The gap between that and a guided capture is easier to see than to describe, so we put the two side by side.
The shape of the fix
This stays broken because every route to fixing it has been expensive to start.
Which makes the useful question a different one. Not how to run a cleanup, since data captured in a campaign starts going stale the day the campaign ends. The question is how to make capture cheap and quick enough that it just happens, in the aisle, at goods-in, by whoever is already holding the item.
Mobile dimensioning is what changes that. Instead of bringing every item to a fixed station, the station goes to the item. The operator measures on a device they already carry, wherever they happen to be standing.
That removes the three things that stall the static route. No capital purchase to specify and approve. No fixed spot that racked stock never passes. No dedicated operator to roster. And because it is software, every value can carry a record of when it was taken and how, which a tape measure will never give you.
That is what Flux does, and it is why TD SYNNEX went that way instead of buying a station.
Downstream, cartonization fed real dimensions instead of estimates takes 15% to 30% of the air out of outbound boxes on our own customers' numbers, which is where the money comes back.
The device is not the point. Get capture cheap enough and the backlog stops growing on day one, which is the only thing that makes clearing it worth doing.
Questions?
It costs money in six places at once: wasted storage space, slower and duplicated picking work, freight reweigh charges and oversize surcharges, under-filled trailers, delayed put-away at the dock, and consultant time on projects that stall waiting for data. No single one is large enough to get funded on its own, which is why the total is rarely calculated.
Check whether any value in the file carries a date and a capture method. Most item masters hold a dimension and a weight and nothing about where either came from. If you cannot tell a laser measurement from a typed guess, the data is functionally out of date whether or not it is wrong, because you have no way to decide what to re-measure first.
Supplier figures usually describe the product or the retail pack, not the shipping unit you actually handle. They also go stale silently: suppliers change packaging, case counts and materials without notifying customers, and nothing in a normal receiving process flags the change.
Aim the cycle rather than the calendar. Re-measure supplier-supplied and estimated values first, leave scanned values alone, and flag anything measured more than a year ago. That only works if every value carries a capture date and a method, which is why provenance fields matter more than a fixed re-measurement interval.
A fixed station needs procurement, installation, an owner and a trained operator before it captures anything, and every one of those steps is a reason to wait. It also only measures what passes the spot where it stands, so stock already in your racking never gets covered. The project is rarely rejected outright, it simply never starts.