Two hundred and forty dormitory beds went into a student residence with a knock-down locker slung under each one. Every door leaf was cut to the factory's cutting list. Every cabinet was cut to a revised list. The two lists never spoke to each other, and the difference between them was twelve millimetres of interference in a rebate that no door could enter.
Illustrative example This case is written from the patterns we see in furniture export orders, and the figures are representative rather than from a specific client file. It shows how a check works and what it finds. We will replace it with a real, anonymised case as soon as we can publish one with the buyer's permission.
Two hundred and forty dormitory beds, and a door that would not shut
A student residence was fitted out with two hundred and forty dormitory bed sets, each with a knock-down storage locker slung under the bed frame. The order was worth forty thousand three hundred and twenty dollars. A deposit of thirty per cent, twelve thousand and ninety-six dollars, was paid, and the balance of twenty-eight thousand two hundred and twenty-four dollars was due against the loading date.
The locker is a simple carcass: two side panels, a top, a bottom, a back, and a hinged door that drops into a six millimetre rebate at the front and closes flush with the carcass edge. What the purchase order specified was a bay of four hundred and fifty millimetres between the bed frame's end uprights, a carcass four hundred and fifty millimetres wide built from eighteen millimetre panels, and a door leaf four hundred and ten millimetres wide.
Those three numbers agree with each other and, taken on their own, they describe a door that fits. A clear opening of four hundred and fourteen millimetres, a two millimetre reveal each side, a four hundred and ten millimetre leaf. Nothing was missing from the drawing. What was missing from the drawing was the only thing that mattered, which was the gap itself, because a gap is never a dimension. It is what is left over after two subtractions, and nobody had written it down.
When the inspector hung a door on the first locker taken from the production run, the leaf would not enter the rebate. It had to be forced, and once it was forced the carcass bowed and the lock would not engage.
This is an illustrative example built from the pattern behind most fit disputes on knock-down furniture: every part is made to its own cutting list, each list is internally correct, and the relationship between the lists exists nowhere except in the assembly. The figures are representative of the category rather than taken from a named customer file.
Where to go next
→ how we check bed and locker sets against your own drawings and tolerances
→ what our reports record, and what they state they could not verify
How the fit was measured
The check was booked as an Order plus Quality Check, two man-days at one hundred and ninety-nine dollars a day. The inspector's first move was not a caliper. It was to assemble a locker from the lot, hang the leaf with the hinges that ship in the bag, and try to close it.
The numbers that came out of the bench and the finished part are all dimensions rather than opinions, and two of them decide the case.
| Measurement | How it was taken | Result |
|---|---|---|
| Leaf width, finished | Across the vertical edges at mid-height, after edging | 410 mm |
| Clear opening in the carcass | Between the inside faces of the side panels, assembled and squared | 398 mm |
| Interference, total | Leaf width less clear opening | 12 mm |
| Interference, each side | Half the total, on an interference fit | 6 mm |
| Reveal at the closed position | Leaf width less clear opening, when it would close | None. It would not close |
| Carcass outer width | Across the side panels, flat | 434 mm |
| Bay in the bed frame | Between the inner faces of the uprights, at the rebate | 440 mm |
Two of those readings decide the case. The leaf was 410 and the opening was 398, so the leaf was twelve millimetres wider than the space it had to occupy and six millimetres too wide on each side. And the carcass was 434 when the order said 450, which is what told the inspector to go and measure the bed frame bay rather than go looking for a cutting error in the wood shop.
Two things about how the leaf was measured are worth stating, because both are ways of measuring the wrong thing. The leaf was measured finished, not cut: the panel goes into the edge bander at 406 and comes out at 410, so a tape on the cut panel understates the finished width by four millimetres. And the leaf was measured in the carcass, not on a bench. A leaf on a bench is a rectangle. A leaf in a rebate is a fit.
Where to go next
→ calculate the sample size and accept numbers for your own lot
→ the eight parts of every verification report, section by section
The number that moved, and the number nobody moved with it
The bed frame's end upright is a folded steel section with a ten millimetre return flange down its inner face, there to stop the section twisting under load. It is on the metal shop's detail drawing and it is on no sales drawing. It is also on every bed frame the factory has ever made, which is why nobody mentioned it.
That flange takes ten millimetres off the bay. The bay was four hundred and forty, not four hundred and fifty, and the locker had to fit into it. The wood shop's answer was to narrow the carcass to four hundred and thirty-four millimetres, leaving three millimetres of clearance each side, and that answer is defensible on its own terms.
The door leaf was a different matter. The leaf cutting list was raised before the bay was confirmed on the floor. When the carcass was revised, the revision travelled on a sheet to the carcass line and never came back to the leaf line. So the carcass was built to 434 and the leaf was cut to 410, and the gap between them was twelve millimetres.
There is no error anywhere in that sentence. The leaf was the right size for the opening the order described. It was the wrong size for the opening the factory built. The number did not go wrong. The number stopped being true, and nothing told the leaf.
Where to go next
→ bed frame and locker verification, measured as assembled rather than as parts
→ desk and table dimensions checked against the contract tolerance
Why the sample passed and the production batch did not
The buyer had approved a first article, and the first article was a good locker. It was built to the sales drawing: carcass 450, opening 414, leaf 410, two millimetre reveal. The door closed. It closed properly, it latched, and it was signed off.
It was also a product the factory was no longer going to build. The flange question was settled on the welding line, where the frames were being made, and the frames were being made after the first article had been approved and after the leaf cutting list had been raised. The approval was genuine and it was worthless, because it described a carcass four hundred and fifty millimetres wide, and the goods that followed were four hundred and thirty-four.
| Approved first article | Production unit | Sales drawing | |
|---|---|---|---|
| Carcass outer width | 450 mm | 434 mm | 450 mm |
| Clear opening | 414 mm | 398 mm | 414 mm |
| Door leaf, finished | 410 mm | 410 mm | 410 mm |
| Reveal each side when closed | 2 mm | Interference, 6 mm | 2 mm |
| Bed frame bay | Not assembled with the frame | 440 mm | 450 mm |
The inspector's note on this is short and it is the most useful line in the file: the sample was not the product. A sample approval tells you the factory can build what you showed them. It tells you nothing about whether the goods in the container came off the same revision, and the only cheap way to find out is to compare the goods against the sample and against the drawing, in that order, and to record the date on each.
Where to go next
→ how we work, and what is outside our scope
→ what a verification report records, including the checks it could not complete
What the report could and could not establish
The report established the fit, and then it established the extent, which is the part that decides the remedy. Calipers on leaves pulled at random from the leaf batch: forty-one of them were 410, and none was anything else. A tape across flat carcasses in the packing area: 434, 434, 434. Bed frames measured at a 440 bay. A systematic error, across the whole lot of 240, on every unit.
What it could not establish was where the number had changed. The factory's cutting records for the revision window were not provided, so the report could say that the deviation existed and was uniform, but not when the carcass width moved from 450 to 434, which document carried it, or whether the leaf list was ever revised and missed.
Two questions were left for the buyer to close, and both are closable with a phone call: the frames already delivered on an earlier order of the same model, and whether the sales drawing the first article was built from was ever revised and reissued. Neither is a question an inspector on site can answer, because both need records the inspector is not given. That is not a gap in the inspection. It is the boundary of it.
Where to go next
→ how we work, and what is outside our scope
→ the five checks a buyer should insist on before the balance moves
Why a bigger sample would not have helped
The AQL table is the right place to send anyone who thinks the answer was a larger sample. The defect was in every one of the 240 lockers, so a sample of one finds it. The plan on the order was general level II on a lot of 240, which is code letter G and a sample of 32, and it would have found this on the first unit pulled.
But the same 32 units behave very differently against a random defect. If three of the 240 leaves had been 410 and the rest 394, that sample would have found it with a probability of about thirty-five per cent. Same lot, same plan, same thirty-two units. The difference is not the sampling. It is whether the error is random or designed in.
The second use of the table is the grading. A door that will not close is a critical defect: the goods do not perform the function they were bought for. Critical is AQL zero, which on this lot gives an acceptance number of zero and a rejection number of one, so a single confirmed locker rejects the lot and stops the container. The inspector reported it that way, and the buyer's instruction to hold the loading followed from the number in the table rather than from an opinion about how bad it looked.
Where to go next
→ calculate the sample size and accept numbers for your own lot
Three ways to fix it, and why two of them fail
The chosen option costs fourteen dollars and twenty cents a leaf in material, cutting and edging, and the site refitting is five dollars and eighty cents a unit. On two hundred and forty units that is three thousand four hundred and eight dollars of leaves and one thousand three hundred and ninety-two dollars of refitting, four thousand eight hundred dollars in total.
The thing that makes this fixable at all is that the containers had not sailed. The leaves were cut on site, the carcasses were flat-packed, and nothing had been assembled yet, so the container could be held, the leaves swapped, and the doors hung and latched on site by the same installers who were going to do it anyway.
| Option | What it involves | Outcome |
|---|---|---|
| Trim 6 mm off each vertical edge of the existing leaf | Edge banding off and back on, hinge and lock bores shifted 6 mm | Rejected. A 398 leaf has zero reveal, so it still binds, and the bores no longer align |
| Add 6 mm battens inside the carcass to open the opening up | Carcass becomes 446 mm outside | Rejected. 446 will not go into a 440 bay, and the carcasses were already packed |
| Re-cut 240 leaves at 394 mm | New leaf, edge banded, hinge and lock bores redrilled, doors hung and latched on site | Taken. 394 gives a 2 mm reveal each side on a 398 opening |
Where to go next
→ what each verification level costs and what it covers
→ send us the order details and we will tell you what a check can cover
The cost of the door, in numbers
The buyer took the rework value out of the balance and paid twenty-three thousand four hundred and twenty-four dollars, and the factory absorbed the cost of the leaves rather than invoicing it. Eleven working days is the part that was not in the order at all. The residence had two hundred and forty rooms to hand over before the academic term, and eleven working days is a fortnight in which those rooms could not be let.
The inspection that found it cost three hundred and ninety-eight dollars. That is the number to hold on to when a buyer is deciding whether to pay for a check on a bed order that is cheap, knock-down, and looks like a box of panels.
| Line | Amount |
|---|---|
| Order value, 240 bed sets | $40,320 |
| Deposit paid at 30 per cent | $12,096 |
| Balance due at loading | $28,224 |
| Inspection, 2 man-days at $199 | $398 |
| 240 replacement leaves at $14.20 | $3,408 |
| Site refitting at $5.80 a unit | $1,392 |
| Total rework | $4,800 |
| Rework as a share of the order | 11.9 per cent |
| Rework cost divided by the inspection cost | 12 times |
| Delay to loading | 11 working days |
Where to go next
→ what each verification level costs and what it covers
→ the five checks a buyer should insist on before the balance moves
What the specification should have said about the opening
The buyer's order was not thin. It named a bay, a carcass width, a panel thickness and a leaf width. What it never did was state the relationship between them, and the relationship is the product.
- State the clear opening as a controlling dimension in its own right, with a tolerance, and say how it is measured: assembled carcass, minimum clear width at mid-height and at the head of the opening, not the width of a panel on a bench.
- State the leaf as a derived dimension, not as an independent number: leaf equals clear opening less 4 mm, a two millimetre reveal each side. Written that way, a change to the opening cannot quietly leave the leaf behind.
- State the bed bay on the same drawing as the locker, as a clear dimension, and require the locker to be trial-fitted into the finished frame before the run. A bay dimension on a sales drawing is a claim; a bay measured on the frame is a fact.
- State the hardware: hinge type, hinge positions relative to the leaf edges, keeper and lock positions. Those are dimensions too, and they are what stop a correctly sized leaf still failing to close.
- State what gets measured and when: the finished, edged leaf, in the assembled carcass, closed, with the reveal recorded at four points, and the result written into the report as a gap rather than as a pass.
- Add a first-article fit clause: one assembled locker with its door hung, closed and latched, and trial-fitted into the frame, approved before production continues. A fit clause costs nothing and runs before the material is committed.
- Say what the gap is and what a failure looks like: two millimetre reveal minimum, and a door that closes and latches under its own weight with no forcing. That sentence on the order would have had the first locker assembled before a single leaf was cut.
Those seven lines are all that was missing. Every one of them would have caught this fault before material was cut, and the last one would have caught it in under an hour.
Where to go next
→ the five checks a buyer should insist on before the balance moves
→ the eight parts of every verification report, section by section
Why the fit, and not the dimension, is what you are buying
Every number in this case was individually correct. The leaf was the size the cutting list said. The carcass was the size the revised instruction said. The bed frame was the size the metal shop had always made. The gap was twelve millimetres, and the gap had never been written on anything.
That is what a dormitory furniture inspection has to be aimed at. Not at the leaf, which was 410 and correct, and not at the carcass, which was 434 and correct, but at the two of them, in the same carcass, with the door shut. A dormitory locker is a set, and a set is verified by assembly rather than by parts on a bench. The report records the gap, the units affected, and the limits of what could be established, and then it says the only sentence that mattered: the door does not close.
Where to go next
→ what our reports record, and what they state they could not verify
→ send us the order details and we will tell you what a check can cover
Have an order you need verified? Send us your PI and factory address. Three levels, published prices, $149 per man-day from.