What is a DFM review?
Design for Manufacture means designing a part so it can be made consistently, at sensible cost, with the fewest problems — matching the geometry, tolerances and material to how the part is actually produced. A DFM review is the point where an engineer who has made parts looks at your drawing and asks whether it describes the part you need, or just the part you drew. Those are different things more often than anyone in this industry likes to admit.
It is not a formality ticked off before quoting. Done properly, it is where you find the missing bore, the radius that doesn't need to be there, the wall that will deform, the tolerance that triples the cost for no reason anyone can name.
Catch those on screen and it costs a conversation. Catch them on the shop floor and it costs a rerun. Catch them at the customer and it costs everything downstream of that. A supply chain rule of thumb says a problem costs roughly ten times more to fix at each stage it survives — the exact multiplier is debatable, the direction is not.
What does a DFM review check?
The physics changes with the process, so a genuine review checks different things depending on how the part is made.
For CNC machined parts:
| Tolerances vs function | Every step tighter than standard costs more, and it climbs fast. A tolerance that matters gets held; a tolerance that doesn't is money spent on nothing. |
|---|---|
| Internal corners | A cutting tool is round — it cannot make a perfectly sharp internal corner. Corners drawn sharp force slower tooling, extra processes, or a design change nobody asked for. |
| Wall thickness | Thin walls chatter, deflect and warp under the cut. A wall that looks fine on screen can be a scrap part on the machine. |
| Tool reach & setups | Deep pockets, features on every face, awkward access. Each can add a setup, and each setup adds cost. Sometimes the fix is one line on the drawing. |
| Threads & engagement | The right thread depth depends on the material. A thread that holds in steel can pull straight out of aluminium. |
For sheet metal and fabricated parts:
| Bend radii & grain direction | Bends have a minimum radius set by material and thickness — and bending across or along the grain changes whether the part cracks. |
|---|---|
| Hole-to-edge & hole-to-bend | Holes too close to an edge or a bend distort when the part is formed. |
| Weld access & distortion | Welding adds heat, and heat moves metal. A joint the torch can't reach, or a weld that pulls the part out of shape, is a design problem, not a workshop problem. |
The interface nobody checks
When a part is both fabricated and machined, the usual failure is treating the two halves as separate jobs. A fabricated frame warps as it is welded. A machined face expects a precise tolerance. Bolt one to the other without thinking about it and the tolerances don't stack up — a weldment and a machined surface are held to completely different standards.
The fix is knowing the sequence and the interface. Precision faces get machined after welding, not before, so the distortion is already in the part when you cut the surface that matters. The joint between a machined face and a fabricated frame gets a slotted hole or a machined pad — not a machining-grade tolerance stamped onto a weldment that will never hold it.
You only design that correctly if you understand both processes. A machining shop treats the fabrication as someone else's problem; a fabrication shop does the same in reverse. Under one roof, the interface is the first thing we look at. It's where the two disciplines argue, and where the part usually fails.
What it looks like in practice
The lever. A stainless steel lever, both fabricated and machined. The drawing specified a precise radius, to be machined in. We asked one question: is that radius functionally critical? It wasn't. So instead of machining it — time on a machine and a feature to inspect — we're testing a different way of producing it, now in sample production. If it proves out, the cost of the part is cut by 45%. None of that came from reading the geometry. It came from asking whether the feature needed to be there at all.
The water tank. A tank in 2 mm aluminium. The drawing didn't specify leak testing and the email didn't ask for it. We put dye testing in the quote anyway — "watertight" can mean a lot of things, and the only way to know a tank doesn't leak is to test that it doesn't. We also flagged the deformation risk: at 2 mm the walls are thin for the design. None of it was requested. All of it is the right thing to offer.
Where the conviction comes from. At Schneider Electric's solar inverter division, our director ran the service department and embedded an experienced service engineer into the new-product team. One early result: screw standardisation. Each design engineer had picked the functionally perfect screw for their subassembly — over two hundred types — and nobody had costed a service engineer travelling to a remote site with no way of knowing which they'd need. Standardising barely touched the product and transformed its serviceability. That is a design decision made better by someone who lives downstream of it. A DFM review is the same principle, pointed at the drawing instead of the field.
DFM review vs posting an RFQ
There's a fast route to a price: upload a file to an instant-quote platform and software analyses the geometry. For a part that is already correct, that's genuinely useful. Here is what it doesn't do: it reads the geometry in the file — it does not ask whether the geometry is right. It can't ask whether your radius is functionally critical, because it doesn't know what the part is for. It won't offer dye testing you didn't request, won't tell you a 2 mm wall will deform, and can't see that your machined face bolts to a fabricated frame. An algorithm reads what is in the file. An engineer asks what is missing from it.
| Post an RFQEngage a DFM review | |
| What gets read | The geometry in the fileThe design intent behind the part |
| Speed | Seconds to a priceA conversation before a price |
| Best for | A part that is already rightA part where the drawing may not be right |
| Wrong tolerances | Sometimes, if the geometry flags itCaught by asking what the part needs |
| Unspecified needs | Never offeredOffered when it's the right thing to do |
| Machined + fabricated | Priced as separate partsReviewed as one assembly at the interface |
For a simple, well-defined part, post the RFQ. For anything where being fit for purpose matters more than being fast to price, have the design reviewed first.
The standards behind the judgment
A DFM review is judgment — but judgment against known standards, not opinion:
- ISO 2768 — general tolerances for machined parts
- ISO 286 — limits and fits
- ASME Y14.5 & ISO 1101 — geometric dimensioning and tolerancing (GD&T)
- ISO 13920 — general tolerances for welded constructions
- DIN 6935 — cold bending of steel
Knowing which standard applies — and spotting where a drawing calls for a tolerance the part doesn't need — is a large part of what the review is for. The other half of the same job is inspection before shipping: the American Society for Quality puts the cost of poor quality at 15–20% of sales for many manufacturers, and most of it is quiet, repeated, avoidable rework.