Shenzhen

Design for manufacture: the conversation that saves the tooling

Between a design that works and a design that can be made repeatably there is a gap, and a factory can usually see it in an afternoon. The whole problem is getting them to say so before the steel is cut.

By RomainFrance → China23.09.2026

Design for manufacture is not a document. It is a conversation between the person who knows what the product must do and the people who know what their machines do badly. It is worth having because the second group holds information the first cannot obtain any other way, and because almost every expensive surprise in a hardware project is a piece of that information arriving too late.

When to ask for it

The window is narrower than people think. Too early and the factory has nothing concrete to react to, so you get generic encouragement. Too late and the remarks are real but unusable, because acting on them means redoing work you have already paid for. The useful moment is when the design is complete enough to be manufactured and not yet committed to tooling: the geometry is settled, the materials are chosen, and nothing has been cut.

There is a second, smaller window that people skip: after the first production run. The factory has now made the part hundreds of times and knows exactly which feature slows the line, which one gets rejected, and which one the operators quietly work around. Asking the question again at that point costs nothing and often produces the most valuable answers of the project.

What the factory knows that you do not

Which of your features is at the limit of what their process holds, rather than comfortably inside it.
Where the mould will have to open, and what that decides about the visible surfaces of your product.
Which tolerance on your drawing will require a slow extra operation that nobody quoted.
Which material you specified is hard to obtain locally, and what they will otherwise substitute.
Which assembly step depends on an operator's care rather than on a fixture — the step that degrades when volume rises.
What they have already had to fix silently on a similar product for someone else.

How to ask so that you get more than « OK, no problem »

« Can you make this? » has one answer, and it is yes. It is not evasion: the question as asked is about capability, the factory is capable, and saying anything else would be turning down work on the basis of a criticism of your design — which, in a commercial relationship where you are the customer, is socially expensive and easy to get wrong. The answer is a rational response to a badly framed question.

So change what you are asking for. Do not ask for a verdict on your design; ask for information about their process. Do not ask whether there is a problem; ask which part is the hardest and why. Put the difficulty on the geometry rather than on the person, give them a way to be helpful rather than critical, and ask in writing so that the engineer — who is rarely the person in the meeting — can answer in their own time and in their own words.

Ask for the answer in Chinese if that is easier for them, and have it translated. An engineer writing in a second language gives you the short version of what they think; the same engineer writing in their own gives you the reasoning, which is the part you actually needed.

Five questions that surface the real problems

Which feature of this part is the hardest for your process, and what would you change if it were your product? It asks for expertise rather than approval, and it gives permission to criticise the design without criticising you.
If you had to reject parts from this drawing, which dimension would fail first? The answer names your real critical dimension, which is often not the one you marked.
Where does the tool open, and where will the marks be? It moves an aesthetic surprise from the first shot to the drawing, while it is still free to move.
What will you do if this material is not available for a batch? Asked before it happens, it is planning. Asked afterwards, it is an argument.
Which step here depends on the operator being careful? That step is where quality will drift when the line is busy, and it is usually fixable with a fixture rather than with a reminder.

What to do with the answers

Not all of them. A factory's suggestions optimise for its own process, which is legitimate and not the same as optimising for your product. Some remarks will make the part cheaper to make and worse to use; some will make it easier for that particular factory and harder to move elsewhere. Sort the answers into three piles: things that are genuinely better, things that are cheaper for them and neutral for you — which are fine, and worth trading for something — and things that quietly change what the product is, which you refuse in writing.

Then write the conclusions back into the drawing, not into an email thread. A design change that lives in a conversation will be implemented by the person who was in that conversation, and forgotten by everyone else the day that person is on another project. The point of the exercise is not to have had a good meeting; it is to arrive at tooling with a design that the people cutting the steel have already read and agreed to.

Questions we get on this.

Should I pay for a DFM review?

Most factories do it for free as part of winning the order, and that is usually enough for a first pass. Paying changes two things: you can ask before choosing the supplier, without the review being a sales document, and you can ask an engineering firm that has no interest in which process you end up using. On a project where the tooling is the largest single commitment, a paid second opinion is cheap relative to what it protects.

The factory says everything is fine. Is that a bad sign?

It is a sign that the question was closed rather than open. Try again with a request for information instead of a verdict: which feature is hardest, which dimension would fail first, where does the tool open. If the answers are still empty after that, the useful conclusion is not that your design is perfect — it is that this supplier is not engaging with it, which is worth knowing before the tooling and not after.

Does DFM mean letting the factory redesign my product?

No, and keeping that boundary is your job. The factory advises on how things are made; you decide what the product is. A useful test on each suggestion: does it change how the part is produced, or does it change what the customer receives? The first is theirs to propose. The second is yours to decide, and it should be answered in writing so that nobody reinterprets it on the line.

Keep reading on this.

ShenzhenTooling: the cost people discover too lateUnit price gets all the attention in a first quote. The mould is what actually locks you to a supplier, and it is rarely discussed until it is paid for.China sourcingWriting a specification a Chinese factory can actually quote againstThe factory is rarely the problem. The specification given to the factory usually is — and it is the cheapest thing in the whole project to fix.China sourcingWriting to a Chinese supplier so that you get a useful answerMost unanswered questions were unanswerable as written. Changing how you ask is faster and cheaper than changing supplier.ShenzhenCatalogue ODM or your own design: what each route costs and what you ownA Chinese factory can sell you a product that already exists with your logo on it, or build the one you have designed. Between the two sits the route most buyers actually take — the modified catalogue product — and it is the one negotiated worst.

If this raised a question about your own project, tell us where you are — the first conversation usually clarifies more than a proposal.

Talk about your product →Read the full guide