The tables were 50 mm short. Under every tolerance the factory works to, they passed.
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.
The situation
A European supplier ordered 40 solid wood dining tables at 1800 × 900 × 750 mm for a hotel restaurant where every table sits in a fixed position against a banquette run.
The purchase order listed the three dimensions. It did not say where to measure them, it did not give a tolerance, and it did not say whether the quoted length was the top or the base frame.
The factory's production drawing carried the same three numbers and nothing more. In the factory's view the order was unambiguous, because there was only one place they had been asked to put a number.
What the inspection found
Units were drawn from across the batch and measured on a flat surface, top face up. Length was taken across the top at both ends and at the centre, and the base frame was measured separately.
| Check | Specification | Measured |
|---|---|---|
| Top length, unit 1 | 1800 mm | **1748 mm** |
| Top length, unit 2 | 1800 mm | **1751 mm** |
| Top length, unit 3 | 1800 mm | **1749 mm** |
| Base frame length | not specified | 1650 mm |
| Width | 900 mm | 896–899 mm |
| Height | 750 mm | 748–750 mm |
Width and height sat inside any tolerance anyone would plausibly write. Only the length was out. That is typical — a shortfall shows up on one axis because one axis is the one with a material-sizing constraint behind it.
Why "industry convention" settles nothing
The factory's position was that ±10 mm is normal working practice in their workshop for a solid wood table. The buyer's position was that a table specified at 1800 mm is 1800 mm. Both positions are sincere, and neither one is a contractual term.
Made-to-order furniture has no dimensional standard in the way a threaded fastener or an electrical component does. There is no published, universally applied tolerance for a solid wood dining table that both parties have agreed to inherit. "Industry practice" is a phrase that gets produced after the fact, and it can be quoted by either side to support the opposite conclusion.
50 mm on 1800 mm is 2.8%. Almost nobody disputes that a dining room can hold a table two percent shorter than ordered. **That is not the argument.** The argument is that nobody can convert that number into a pass or a fail, so the decision stops being a measurement and becomes a negotiation — conducted after the container is sealed and the demurrage clock has started.
The only document that can settle this is the purchase order. Where it carries no tolerance, there is no non-conformance to report. That is a worse outcome than a defect, because a defect can be fixed and an absence of a specification can only be renegotiated.
The report position
The report did not state that the tables were non-conforming. It could not, for the reason above. What it recorded was the specification exactly as written, the measurement method, the readings, and a note that **the base frame length had no corresponding figure in the purchase order at all**.
That second observation is the one that moved the conversation. It established that the order did not fully describe the product being bought — the 150 mm difference between top and frame is the designed overhang, and it appeared nowhere in the specification. Both parties had been working from an incomplete description and treating it as complete.
The factory's explanation was a cutting allowance carried over from an earlier revision of the design, when the top was a single board rather than a jointed two-board construction. Whether that accounts for the full 50 mm was never established. The cutting list was requested and not produced before the container left.
**The report said what was measured and stopped there.** It did not attribute a cause, and it did not use the word *defect*. Both would have been conclusions the data could not carry.
What to put in the specification
One more line is worth adding whenever tables are supplied against a fixed layout: **confirm the top length against the installation drawing before production starts, not after delivery.** A 50 mm shortfall on a free-standing table is an irritation. On a run of banquette seating it is forty tables that do not line up with the joinery, and the joinery has usually been fabricated already.
- **The nominal value and the tolerance together.** For a solid wood top, ±5 mm is a realistic working tolerance. Panel and veneer tops hold tighter. ±10 mm on an 1800 mm top is 0.55% and nothing structural depends on it — but 50 mm is 2.8%, and the number belongs in the document rather than in the argument.
- **Which part of the product the number applies to.** A table has at least two lengths — the top and the base frame — and they differ by the overhang. "1800 mm" on its own does not say which one was ordered.
- **Where the measurement is taken.** Top face up on a flat surface, read across the top at both ends and at the centre. Not with a tape pulled around a profiled or bullnose edge, which reads long and will hand the factory an argument for free.
- **How many units are measured, and what happens outside tolerance.** A sample size, and the remedy — rework, replacement, or a stated concession accepted in writing. Without the last part, a tolerance is only a record of the problem.
Where to go next
→ dining table and chair verification in our dining scope
→ what a report records when there is no specification to measure against
Have an order you need verified? Send us your PI and factory address. Three levels, published prices, $149 per man-day from.