A hotel supply company ordered 450 units of bedroom furniture for twelve properties. The verification report closed with two sentences: release recommended, goods conform to the purchase order. The buyer attached that PDF to a retail listing as the quality evidence. Six weeks later the retailer's compliance team asked for the accredited test report the listing implied, and there was not one, because the same report said on its last page that it was not a certificate.
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.
A PDF attached to a product listing
A hotel supply company was fitting out twelve properties and placed one order for 450 units of bedroom furniture: 300 upholstered guest-room accent chairs and 150 solid oak bedside tables. The order ran to eleven lines. Models, quantities, dimensions, timber species, fabric composition, carton marks, delivery window, and a closing line asking for pre-shipment inspection by an independent third party.
The inspection was booked at the order and quality level, two man-days, with the loading slot at the end of the visit. The report came back as nine pages and a photograph folder, issued within a day. The buyer read it once, read the conclusion, and forwarded the PDF to the retailer whose shelves the furniture was going on.
The retailer listed the product with the report attached as the quality evidence, because that is what every other listing on the platform carried. Two of the twelve properties were already installed when the listing was questioned.
This is an illustrative example built from the patterns behind most inspection report disputes on furniture orders: the inspection was competent, the report was honest, and the reader took the conclusion to be something it was not. The figures are representative of the category rather than taken from a named customer file.
Where to go next
→ the eight parts of every verification report, section by section
The question that found it
It did not come from the furniture. It came from the retailer's compliance team, six weeks after the listing went up, asking for the test report behind the claim. Specifically they asked which accredited laboratory produced the formaldehyde result, and whether the verification report had been issued under CMA or CNAS.
The buyer's procurement lead forwarded the question to the inspection company. The answer was that the report was not an accredited laboratory report, it was a verification report, it carried no legal standing, and it could not be used for marketplace approval. That sentence was in the report the whole time. It was in the footer of page nine, under the pricing line, in nine-point type.
Nobody had read page nine. The furniture was fine. The claim made on the strength of the report was not, and the two are not the same thing.
Where to go next
→ how an order verification runs, from booking to the written report
Eight sections, and a different reading order
A verification report has eight sections, and they are written to be read in order, because that is the order the evidence was gathered in. Buyers do not read them in that order. They read the recommendation, which is section seven, and stop.
The order that protects you is the reverse of the order that is convenient. Read section eight first, because it caps what every other section is allowed to mean. Read section two next, because it defines the yardstick the conclusions were measured against. Only then read section seven, which will by then mean something.
There is a reason this works, and it is not a stylistic preference. In the accreditation system for test and calibration laboratories, the requirement to report conformity separately from results is explicit, and the requirement to state that results apply only to the items sampled is explicit. Both requirements exist because readers routinely take a report's conclusion as a statement about the whole lot, and about legal compliance, when it is neither.
Any inspection report sample the factory or the inspection company sends you will have the same shape, and usually the same weakness in the footer. You do not need a consultant to tell you which two sections to read first. You need to read them first.
Where to go next
→ what our report records, and the checks it could not complete
Section two is the yardstick
Section two lists the documents the inspection was judged against. In this case there were four: the purchase invoice, the packing list, a photograph of the fabric swatch, and an email from the buyer's merchandiser confirming the colour. The order had eleven lines. Seven of them were not in the file.
That is not the inspector's fault, and section two says so plainly. It also means that the sentence in section seven, goods conform to the purchase order, was a statement about four documents. It was true. It was also much smaller than it sounded.
| What the order said | What the inspector had | What could be checked |
|---|---|---|
| Eleven lines: models, quantities, dimensions, species, fabric composition, carton marks, delivery window | The purchase invoice | Model and quantity conformity, carton marks |
| A dimension table with limits | Nothing | Nothing. Measurements were recorded, not graded |
| Fabric composition and colour against a physical approved sample | A photograph of a swatch | Nothing comparable. A photograph cannot be held next to the cloth |
| A material compliance requirement, with a named standard and a limit | Nothing | Nothing. The requirement was in nobody's document |
| A packaging specification and a drop test protocol | Nothing | Carton count and visible condition only |
Where to go next
Section eight is where the report protects you
Section eight lists what was not checked, and why. This report had five entries. Three of them were things the buyer had assumed were covered, and one of them was the reason the listing was questioned.
The reasons fall into three kinds, and the kind matters. A reference was missing. The check was outside our scope. Or the check was beyond what we can do on site. The first kind is the buyer's to fix. The second and third are the buyer's to budget for. None of the three is a criticism of the factory.
- Formaldehyde: no accredited laboratory report was supplied with the order, and an on-site inspection cannot produce one. Not verified. The buyer's supply chain manager had assumed a furniture inspection included an emissions test, because for their previous supplier it did, at a different supplier, at a different price, in a different country.
- Timber species on the bedside tables: checked visually against the order wording and against the mill certificate the factory supplied. The report recorded it as a document check, not a species determination, and a document check is not a laboratory result.
- Seat height and overall dimensions: measured on the sampled units, but no tolerance schedule was supplied, so the numbers were recorded and not graded. A measurement without a limit cannot be a pass or a fail.
- Drop test on the cartons: not performed. No protocol was specified by the buyer, so there was nothing to test against.
- Assembly and function: checked on the sampled units only. The report recorded that the result applies to the units sampled and not to the remainder of the lot.
Where to go next
The sentence about the sample that was missing
ISO/IEC 17025 clause 7.8.2.1 requires every test and calibration report to carry a statement that the results relate only to the items tested, calibrated or sampled. Accreditation bodies have been explicit about why. European Accreditation has answered that this statement is needed so that results cannot be extended to other items, and cannot be read as lot approval, product certification, or an inspection activity.
This verification report did carry the sampling basis, which is the count side of it: the lot size, the sample size, and the accept and reject numbers applied. The chairs were 300 units, code letter H, 50 opened. The tables were 150 units, code letter F, 20 opened. Seventy units were examined out of 450. Two minor defects, inside the acceptance numbers, nothing hiding in the counts.
What the report did not do was carry the limit into the sentence a reader remembers. Release recommended, goods conform to the purchase order, reads as a statement about 450 units. It was a statement about 70. The sampling basis is a number in a section; the conclusion is a sentence, and sentences travel. That gap is where the buyer of this order got it wrong, and a report closes it in one line: conform to the purchase order on the 70 units sampled, listed in section three.
Where to go next
A recommendation is an opinion, not a conformity statement
The same accreditation system draws a line that most furniture reports never draw. A conformity statement is a technical statement. Clause 7.8.6 requires it to identify which results it applies to, which specification or standard is met or not met, and which decision rule was applied. An opinion or interpretation, under clause 7.8.7, is different in kind. It is a judgement. It has to be based on the results, it has to be identified as an opinion, and whoever issues it has to be authorised to do so.
Release recommended is an opinion. It is a competent opinion, and on this order it was probably the right one. It is not a conformity statement and it is not a legal finding, and the buyer who forwards it as one has quietly changed what it says.
| As written | As a conformity statement reads |
|---|---|
| Release recommended. Goods conform to the purchase order. | Conforms to PI lines 1 to 4, on the 70 units sampled and listed in section 3. Not graded against dimensions: no tolerance schedule supplied. |
| Seat height 462 mm. | Seat height 460 to 464 mm on 20 units. Limit not supplied by the buyer, so recorded and not graded. |
| Timber species consistent with the order. | Factory mill certificate sighted for the oak. No species determination performed by the inspector. |
Where to go next
→ the findings and recommendation section of a verification report
→ why we would rather lose a fee than sign a report claiming a check we did not make
The decision rule nobody wrote down
Conformity needs a rule, and this is the least glamorous part of any report and the most useful. ILAC-G8:09/2019, the guidelines on decision rules, sets out what a laboratory has to state when it says pass or fail. Under simple acceptance, where the acceptance limit equals the tolerance limit and no guard band is applied, the probability of falsely accepting a result sitting at the limit can be as high as 50 per cent. With a guard band equal to the expanded measurement uncertainty, the false accept and false reject risk on a single result falls below 2.5 per cent.
Furniture inspection lives at the coarse end of this. A tape measure against a seat height carries very little uncertainty, so simple acceptance is usually the honest choice. The point is not to argue for guard bands on a bedside table. The point is that a report which reads 448.3 mm against a requirement of 450 plus or minus 2 mm, and passes it, has told you nothing about whether the pass came from a rule or from a rounding.
One bedside table leg measured 448.3 mm against 450 plus or minus 2 mm. Simple acceptance passes it. The same reading with a stated guard band of 0.8 mm would fail, because the acceptance limit would sit at 448.8 mm. Whether that unit matters is the buyer's decision, not the inspector's, and a buyer cannot make the decision from a report that never says which rule was applied.
Where to go next
→ what each verification level costs and how many man-days it takes
The requirement that was never in the order
The formaldehyde obligation on the listing side did not come from the furniture order. It came from regulation. REACH Annex XVII Entry 77 sets a limit of 0.062 milligrams per cubic metre for furniture and wood-based articles, and it has applied since 6 August 2026. EN 717-1, the one cubic metre chamber method, is the most straightforward route to demonstrate it for furniture. EN 16516 is the method most European buyers specify, and results from it can be used directly only where the measured value sits at or below the legal limit, with a validated correlation needed to go above it.
Nothing in this buyer's order mentioned formaldehyde, an emissions standard, a limit, or a laboratory. So the inspector could not check it, and section eight said so in one line. Nothing in the order mentioned it because it was assumed: the retailer had said the listings need compliance evidence, the buyer had assumed a furniture inspection was that evidence, and nobody wrote it down.
This is the same gap as the tolerance schedule and the approved sample, in different clothes. A requirement nobody put in the order cannot be checked, cannot be graded, and cannot be defended later. The inspector did the right thing by writing it into section eight. The buyer nearly lost a listing over a line they had skipped.
Where to go next
How the report should have been written
Nothing about the inspection was wrong. The report was honest, the findings were correct, the recommendation was reasonable, and the limitations were all disclosed. It was written for the inspector's order of work rather than the reader's, and the reader's order is different.
Four changes to the writing would have prevented everything that followed. Carry the scope limit into the conclusion. Name the criteria the conclusion is about. State the decision rule next to any graded result. And put the standing of the document on page one rather than in the footer of page nine, because the question it answers, what is this and what may I use it for, is the first question a buyer has and the last one they should have to go looking for.
| As written | As it should read |
|---|---|
| Release recommended. | Release recommended. On the 70 units sampled and listed in section 3. The remainder of the lot is not covered by this report. |
| Goods conform to the purchase order. | Conforms to PI lines 1 to 4, the four documents supplied. The order's other seven lines were not supplied and could not be checked. See section 8. |
| Not a product test report. No legal standing. | Page one, under the title: this is a verification report, not an accredited laboratory report, and it is not evidence of regulatory compliance. Section 8 lists what was not checked. |
| Measured 462 mm. | Seat height 460 to 464 mm, 20 units, not graded: no limit supplied by the buyer. |
Where to go next
What goes into the next order
The fix for this order was not a better inspection. It was a better order, and the list is the same every time: the requirements that decide whether the goods can be sold, named one by one, with the document that proves each one.
- Name the material compliance requirement, with the standard, the limit and the method: a chamber method to EN 717-1 at 0.062 milligrams per cubic metre for the upholstered and composite wood components, plus which laboratory the report must come from and who pays for it. If the accredited report is a document to be collected, say that it is.
- Require the accredited report as a separate deliverable before release, and say in the order that the verification report does not replace it.
- Send a tolerance schedule. Every dimension on the order with a figure and a limit, so measurements can be graded instead of merely recorded.
- Send a physical approved sample and keep a copy. Colour and fabric compared against an object, not against a photograph on a screen.
- State the sampling basis in your own words: whether the order is one lot or one lot per production week, and the inspection level and AQL you want applied, so the report's numbers mean what you think they mean.
- Ask for the report to state its scope limit in the conclusion, and to name the decision rule used for any borderline measurement. Two lines, and they are the two lines this case needed.
Where to go next
→ the five checks a buyer should insist on before the balance moves
Two questions before you read a single finding
The buyer recovered the listing by commissioning the accredited test themselves, at 620 US dollars and eleven working days, and by taking the claim down until the report arrived. The furniture never moved. The eleven days did.
So two questions, asked before you read anything: what were you judging against, and what could you not judge. If the answers are on page one, you are reading a well-made report. If they are on page nine, they are still on page nine, and the only thing you can do with them is decide whether they matter to you.
Where to go next
→ 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.