12 Questions to Ask a Jewelry Manufacturer Before Your First Order

A first order with a new workshop is not really a purchase — it is a test of how that workshop works. The price is the least informative number in a first quote. These twelve questions separate a supplier who has read your design from one who has only read your inquiry — and the table below is the version you can keep next to the email.

The twelve questions on one page

Each question has an evidence to ask for, not just an answer to listen to. If a supplier cannot produce the evidence, the answer is an opinion.

Why the questions matter more than the price

A first quote prices an object. What you are actually buying is a route: how the design gets divided, what gets finished by hand, which joints are closed last, and what happens when a piece fails inspection on a Tuesday afternoon. None of that shows up in the number. It shows up in the answers to the questions below — and a supplier who cannot answer them has usually not looked at your files as a manufacturing problem yet.

Ask these before the first order, not after the first sample. Every one of them is cheaper to ask now than to discover later.

About the route — questions 1 to 4

1. “Have you reviewed my files as a manufacturing problem — and what did you change?”

A workshop that quotes without touching the design is selling you a price. A workshop that comes back with a marked-up drawing — a seat depth adjusted, a wall thickened, a section line proposed — is telling you they intend to build it. Changes proposed before an order are the cheapest changes you will ever get. The features that fail on the bench are exactly what this mark-up should catch.

2. “Where would you split this design, and why?”

Even for a design that can be cast in one piece, this question has a good answer. It tells you whether the supplier thinks in sections, whether they know which stone seats are trapped inside the geometry, and whether the split they propose is the lowest number of sections the design allows. If the answer is “we would cast it whole” for a piece with piercings and interior seats, that is not confidence — it is a routing decision nobody has examined. The fabrication versus casting comparison is the vocabulary to have this conversation in.

3. “What determines the quote — metal weight, bench hours, or route?”

All three are legitimate. What matters is knowing which one dominates, because it tells you where the risk sits. A quote driven by metal weight moves with the market; a quote driven by bench hours reflects how the piece is actually made; a quote driven by route reflects a supplier who has thought about the build. A supplier who cannot tell you which of the three their number is built on has not built it yet.

4. “What in my files would you flag before starting?”

Every design has a soft spot: a tolerance too tight for the material, a finish specified where the tool cannot reach, a stone size the seat geometry will not hold. A good supplier names theirs in the first reply. “Everything looks fine” on a first read usually means the read was fast.

About the sample gate — questions 5 to 8

5. “What exactly does the approved sample lock in?”

A sample approves something specific: dimensions, metal and stone specification, finish, weight within a tolerance. Ask which of these the sample is the reference for, and how it is recorded. A sample that locks in everything is your best asset in the whole project — ask what will still be allowed to move afterwards, and get the answer as a list rather than a nod.

6. “Who signs off the sample, and against what?”

The answer should be a comparison, not an opinion. The sample is checked against the drawing, the specification, and the agreed tolerances — by someone whose job is to find the difference between them. Ask what happens when the check finds one.

7. “If the sample and the drawing disagree, which one governs?”

This sounds like a lawyer's question. It is a manufacturing question. The drawing is what you asked for; the sample is what you approved. When they differ — and on a first order they often do — the answer decides whether your reorder matches your first run or your second thoughts. Get the answer in writing before the sample is made.

8. “What changes after approval, and what does each change cost?”

Changes are normal. What is not normal is finding out after the fact that a change you considered small reopens a whole stage. Ask which changes are absorbed and which are not — the pattern of the answer tells you how the stages are actually sequenced. Tying each payment to a stage makes this question much easier to enforce.

About the run — questions 9 to 12

9. “How is consistency maintained across the run?”

Not “do you check quality” — everyone says yes. Ask what gets measured, at which point in the sequence, and against what reference. In a continuous small batch run, consistency comes from checkpoints at fixed points, not from one final inspection at the end.

10. “What happens when a piece fails inspection mid-run?”

Every run produces a failure somewhere. The question is whether the failure stops the line and gets diagnosed, or gets quietly absorbed into the delivery count. A supplier with a real answer will describe a gate, not a guarantee — and will tell you who decides whether a borderline piece ships.

11. “How are my files handled — who sees them, and what happens to them afterwards?”

Drawings, CAD files and samples are your project material. Ask who on the workshop side has access during development, and what happens to the files when the project closes. The mechanics matter more than the promise — staged file release and a named team answer more questions than an NDA signature. Our own handling is described in the confidentiality note and the privacy policy; expect the same clarity from any supplier you consider.

12. “If we reorder in a year, what makes the second run match the first?”

Repetition is a process question, not a memory question. The answer should involve a recorded reference — the approved sample, the routing decisions, the fixture set — not “we keep your files.” Files tell you the shape; they do not tell you which joints were welded in what order or which surfaces were finished by hand. That record is what makes a reorder a reorder. The cost of losing that record is what you pay when you switch.

How to ask twelve questions without sending twelve questions

Sending all twelve at once gets you a polite reply to four and silence on the rest. Ask them in three rounds, matched to the decision you are actually making. Each round is short enough that a workshop can answer it properly, and each answer tells you whether the next round is worth sending.

1. Round one, with the inquiry: questions 1, 3 and 4. Send the drawing with the specification. You are testing whether anyone reads it before pricing. A reply that quotes a number and nothing else has answered the round.
2. Round two, before the sample: questions 2, 5 and 7. Now you are deciding the route and the reference. This is where the split, the sample lock and the conflict rule get settled — while changing them is still free.
3. Round three, before the run: questions 6, 8, 9, 10 and 12. The gate questions. Ask them once the sample exists, because now the answers can be checked against a physical object rather than a description.

Question 11 — file handling — belongs in whichever round your purchasing process needs it. If you cannot sign an NDA before sharing drawings, ask it before round one and expect a mechanism, not a reassurance.

Three answers that come up every time

“Isn't a list of twelve questions rude?”

No — a numbered list is the easiest email to answer. What is rude is a vague inquiry that forces the supplier to guess, and then a complaint that they guessed wrong. Send the questions with your files and say plainly which three matter most to you; a good workshop will answer those three first.

“What if they only reply with a price?”

Then you have learned what you needed to know, cheaply. A price-only reply is not proof of a bad workshop — it may be a busy one — but it is proof that nobody has looked at your piece yet. Reply once with three specific questions. If the second reply is still a number, the third reply is unlikely to be a route.

“Should I ask to see similar work?”

Yes, and ask a better question than “can I see your portfolio”: ask which piece of similar work gave them the most trouble, and what they changed because of it. A workshop that can answer that has a memory of its own production. A gallery of finished photographs cannot tell you that, which is why reading the answers matters as much as asking the questions.