What gets lost each time a drawing changes hands
An instruction to cut stone passes through six pairs of hands. What goes missing at each one, and three things to put on any drawing you send.
10 November 2026 · 15 min read · GranTerra Stone, Hui'an
An instruction to cut stone travels further than most people expect. It begins as something a designer or a client has in mind, becomes a drawing, becomes an email, becomes a question from a workshop, becomes a cutting list, becomes a number on a crate, and finally becomes a piece being carried up a staircase by two people who have never seen any of the earlier versions. At every one of those steps something is dropped, and it is almost never a fact. It is the background that made the fact mean what it meant.
The reason this matters commercially is that the loss is invisible until it is expensive. Nobody sends an email saying they have removed a piece of context. The document still looks complete. The instruction still reads as an instruction. What has gone is the part that told the reader which way to understand it.
This is not a language problem, although it looks like one from the outside. It is a problem of reference. Two people can share a language perfectly and still fail, because they are picturing different objects.
The same drawing can be read three different ways
Hand a single sheet to three people and you will get three readings. The designer reads it as an intention: this is the effect I want, and the lines are how to get there. The person quoting reads it as a quantity: how many pieces, what area, what work. The workshop reads it as a sequence of operations: which face is finished, which edge is profiled, which end is cut first.
None of the three is wrong, and none of them is the whole document. The trouble starts when one reading is assumed to be the only one. A workshop that reads only as operations will produce exactly the drawing and miss the effect. A designer who reads only as intention will not notice that a detail cannot be made. A quoter who reads only as quantity will price a thing that is not what was drawn.
This is why the same drawing produces quotes that differ by a wide margin, which is the subject of comparing three prices against one drawing. Some of that spread is margin. A good part of it is three people reading three different documents that happen to look the same.
What is lost at each handover is the assumption, not the number
A dimension travels well. A number written on a drawing arrives intact, because there is nothing in it to interpret. What does not travel is everything that was obvious to the person who wrote it.
The obvious things are, in practice, the ones that matter most. That this wall is out of square by more than the drawing suggests. That the pattern has to run this way because of where the light comes from. That the client has already rejected one version of this detail. That the piece is going into a space with a low ceiling, so the largest single unit that can be carried in is smaller than the drawing implies.
None of that belongs on a cutting list. All of it changes the cutting list. And the further the drawing travels from the person who knew it, the fewer of those facts survive, until eventually the document arrives at a saw with nothing attached to it except its own lines.
The practical consequence is that a drawing should never travel alone. It should travel with the two or three sentences that a workshop cannot derive from it: which face is the front, which way the pattern runs, and what the piece has to fit against when it arrives. Those sentences take a minute to write and they are the difference between a batch that is right and a batch that is exactly as drawn and wrong. The mechanics of turning a drawing into something a saw can follow are set out in cutting to your drawing.
A photograph without a scale cannot be argued with
This is the most common failure in the whole chain, and the easiest to fix. A photograph arrives showing a chip, a colour difference, or a piece that does not look like the sample. It is examined, discussed, and acted on. Then somebody asks how big the thing in the picture is, and there is no answer, because the photograph contains nothing of known size.
Without a reference, a camera does whatever it wants with scale. A chip photographed from close range looks like a fracture. A variation photographed under a work light looks like a defect. A perfectly normal slab photographed at an angle looks bowed. Every one of those judgements is a real judgement made by a real person, and every one of them is unsupported by the image.
The fix is not better photography. It is putting something in the frame whose size is not in question. A tape measure is the obvious choice, and it is also the clearest, because the numbers are readable and nobody has to guess. A hand, a coin, a pencil or the corner of a sheet of paper also work, less precisely. What matters is that the reader can convert what they are looking at into something they know.
The same principle applies to colour, where the problem is worse because the camera decides it. A stone photographed under a warm bulb and the same stone photographed outdoors are two different materials as far as the image is concerned. Judging a surface from a picture needs something neutral in the frame, and the note on reading a slab deals with what else the image has to show before it is worth an opinion.
Which face is the front is almost never obvious
Stone has a face and a back, and they are not interchangeable. The face is finished, and the back is not. On a veined material the face also has a direction, and the same piece turned the other way is a different visual object. A drawing that shows an outline without saying which side is which has specified a shape and left the material undecided.
This becomes acute wherever the edges are worked. A profiled edge, a chamfer, a bullnose — each is cut on a specific arris, and cutting it on the opposite one produces a piece that is dimensionally correct and visually wrong. The workshop can see the profile in the drawing. It cannot see which face the designer meant, because the drawing does not say.
The same applies to veined and figured stone in a way that has no equivalent in a plain material. Two adjacent pieces from the same block, laid with the figure running opposite ways, read as a mistake to anyone looking at the finished surface. Both were cut correctly. Neither was cut wrong. The instruction simply never mentioned direction, and direction is not a detail that a saw can infer.
Curved work concentrates this problem, because a curve has a front and a back in a way that a rectangle does not, and the piece that looks right on the drawing can be the mirror of what was wanted. The note on shaping a curve from a block sets out why the template has to carry that information rather than the drawing alone.
“Like the one at the hotel” is not a reference
A reference only works if the person receiving it can see the same thing. A phrase that points at a building, a previous project, or a photograph from a brochure fails for a simple reason: it assumes shared access to an object that only one party can actually look at.
What arrives in the workshop is a description of a memory. The person writing it remembers a warm grey with a bit of movement. The person reading it has no idea whether that means a fine even grain or a strongly figured stone, and will make a reasonable choice based on the words, which is a different choice.
The substitute is not more description. It is an object. A physical sample, sent or held, ends the argument before it starts, because both parties are then referring to the same piece of material rather than to two images of two different pieces. Where an object cannot travel, the next best thing is a photograph of the actual intended piece with something in frame for scale and colour, which is why the sample you keep matters more than the sample you send.
This is worth being firm about, because the failure is expensive in a specific way. A wrong material discovered at the sample stage costs a conversation. The same wrong material discovered after the batch is cut costs the batch, and the note on deciding what counts as acceptable exists because at that point the argument has moved from what was wanted to what is defensible.
A voice message cannot be forwarded to a foreman
Voice messages are efficient and they feel warmer than text, which is why they get used for exactly the questions that most need a record. A thirty-second recording can settle a detail in less time than it takes to type it, and it does so by carrying tone, hesitation and emphasis — all the things writing is bad at.
It also does something else. It creates a piece of information that exists in one place, on one device, addressed to one person. It cannot be attached to a drawing. It cannot be read by somebody who was not in the conversation. It cannot be searched, and it cannot be produced six months later when the question of what was agreed comes up.
The habit that works is to use voice for the discussion and text for the conclusion. Talk it through if that is faster. Then write one short message that says what was decided, in sentences, and let that be the version that travels. The recording did its job by getting to the answer. The message does the different job of making the answer portable.
There is a second, subtler cost. A voice message is delivered in one language, at one speed, to one listener. As soon as it has to be explained to a third person — a foreman, an installer, a colleague covering for someone on leave — it is being paraphrased, and paraphrase is where the original meaning quietly changes.
The questions that come back before cutting starts
On any drawing of ours that goes to the workshop, a small set of questions tends to come back, and they come back for the same reason every time: the drawing is complete as a shape and incomplete as an instruction.
Which face is finished, and which is the back. Which way the figure runs, in the case of a material that has one. Whether the pieces are numbered to a position or interchangeable. Whether the edges are worked before or after the size is established, because those two sequences give different results if anything moves. And what the piece has to fit against on site, which is the question nobody puts on a drawing and everybody needs.
These are not difficult questions and answering them takes minutes. What they do is move the decision from the workshop, where it is made by whoever is at the bench that morning, to the drawing, where it is made once and stays made. The difference in outcome is not subtle, and it is the reason a cutting list is worth more than a pile of drawings with the same information spread across them.
The ambiguity nobody knows to ask about
Here is the part that cannot be fixed by being more careful. We ask about the ambiguities we can see. We cannot ask about the ones that do not exist on our side of the water.
A drawing that shows a straight run of stone against a wall assumes the wall is straight. If it is not, the error is on site, and it will be discovered by the person fitting the last piece rather than by the person reading the drawing. A drawing that specifies a finished floor level assumes the substrate is where the drawing says. A drawing that shows a lift assumes the lift will take the crate.
None of these are stone questions, and none of them are visible in a document produced by a stone supplier. They are the reason a handover is not a one-way transfer, and the reason a short description of the building is worth attaching to a drawing even when nobody asked for it: how it is accessed, what it sits on, and what it has to meet at its edges. Stone rarely meets only stone, and the junctions with other materials are where most of those edges get settled.
The general rule is uncomfortable but useful. Every assumption in a document is invisible to the person who holds it, and visible to the person who receives it, and neither party can see the other’s set. The only way to close that gap is to say out loud what you are taking for granted, which is a strange thing to do in a commercial email and is exactly what prevents most of the expensive mistakes.
Write it where it will be read by someone who was not there
The test for any instruction is not whether the person you are writing to will understand it. It is whether a person who joins the project in three weeks, with no access to the conversation, will understand it.
That reader is not hypothetical. They are the foreman who starts on Monday, the installer who arrives after the stone, the colleague covering a holiday, the person at the bench on the morning the drawing is finally cut. In a project that runs for months across several time zones, almost every reader is that reader.
Writing for them changes the shape of what gets written. Sentences get shorter and more specific. Pronouns get replaced with names. A reference to “the earlier version” becomes a date and a file. The result is a document that is slightly longer and much harder to misread, and the effort is front-loaded in exactly the place where it is cheapest.
It also changes what gets kept. A conclusion written in a message that names the piece and the decision is a record. The same conclusion discussed on a call is a memory, and memories diverge. This is the same reason the final handover has to be a document rather than a conversation, which is the argument in what a signature at the end of a job actually proves. What that pack contains is fixed, but which of its parts belong to the factory depends on how far the supply runs.
Three things to put on any drawing you send
The first is direction. An arrow, a word, a note saying which edge is the top and which face is the front. It costs nothing and it removes the largest single category of error in shaped and veined work, because it converts an inference into an instruction.
The second is a reference of known size, in any photograph that accompanies the drawing and in any photograph sent later to discuss a defect. The tape measure in the frame does more for a claim than a paragraph of description, and it does it in both directions: it supports a legitimate complaint and it defuses an exaggerated one.
The third is position. Not just how many pieces, but where each one is going, so that a piece can be traced from a number on a crate to a place on a wall. That is what makes a set into a set rather than a quantity, and it is why the numbering habit described in putting a number on a crate has to start at the drawing rather than at the packing bench.
Two of those three are things a supplier can ask for and cannot supply. Direction and position we can propose, and often do, because a drawing arrives without them more often than not. The reference of known size is the one only the person on site can provide, and it is the single most useful thing anyone sends us. A drawing, a photograph with something measurable in it, and one line saying what the piece has to fit against — with those three, most of the distance between an idea and a stone disappears. The rest is a conversation about the details, and details are the easy part once the reference is shared. It can start at the contact page with whatever exists so far, because a rough sketch on a sheet of paper is a better beginning than a description polished for a week.
Every figure on this page comes from our own order sheets. Where something is a trade norm rather than a number we can hold, we say so — send a spec and we will confirm it against a real block before you commit.
More from the yard
Send the drawing. We do the rest.
Reply within one working day, in English or Chinese. Send a spec, a drawing or just a photo of the sample you are matching — we will tell you honestly whether we can hold it.
MOQ 1 × 20FT · FOB XIAMEN · SAMPLES IN 7 DAYS