Machining cost is not a single number to negotiate down. It is the sum of a set of decisions, most of which were made before the RFQ was sent — and the largest savings usually come from changing the design or the specification, not from pressing a supplier on rate.
Here are seven levers, roughly in order of impact, with an honest note on what each costs you.
This is the single biggest lever and the most common waste. A general tolerance of ±0.005 mm across an entire part will add 40–100% to the price, and most of those dimensions will never be measured.
What to do: put an honest general tolerance (±0.1 mm for most commercial parts) in the title block, and mark only the dimensions that carry function. On a typical bracket, that alone moves 60–80% of dimensions from precision to commercial.
What it costs: nothing, if the analysis is right. It requires you to know which dimensions matter, which is work — but it is design work, not manufacturing cost.
Every face that must be re-fixtured is a setup, and every setup carries programming, fixturing, indicating and inspection. A part that goes from four setups to two often drops 25–35% in price without changing a dimension.
What to do: design so that as many features as possible are reachable from one direction. If five faces need features, ask whether a 5-axis machine running one setup is cheaper than a 3-axis machine running four — at low quantity it often is, despite the higher hourly rate.
What it costs: possibly a design change, and 5-axis may limit which suppliers can quote.
Deep pockets force long, slender tools that deflect and chatter, which forces light passes and long cycle times. Internal corner radii smaller than the pocket depth require ever-smaller cutters with ever-lower material removal rates.
What to do: keep pocket depth below about 3× the cutter diameter where possible, and set internal corner radii at one third of pocket depth or larger. If a sharp internal corner is functionally required, expect EDM as a separate operation with its own cost and lead time.
What it costs: a fillet where you wanted a corner — usually invisible in function, occasionally important for stress concentration (where a fillet is actually better).
Material cost is often less than the machining cost it causes. Titanium is 5–8× the price of aluminum per kilogram and machines at one third the speed with far shorter tool life — the part ends up 3–6× the aluminum price.
What to do: check whether 6061-T6 or a 400-series stainless meets the actual requirement before specifying titanium or 17-4 PH. Where high strength is needed in a small area, consider a localized insert or a different heat treatment rather than upgrading the whole part.
What it costs: engineering judgement about the real load case. Occasionally the answer is genuinely "it must be titanium".
A part designed around a 1.8 mm sheet will be made from 2.0 mm or 1.5 mm, and either way it is not the part you modelled. A part whose envelope forces a custom plate size carries the whole plate's cost.
What to do: pick from standard stock: 1.0/1.2/1.5/2.0/2.5/3.0 mm sheet, standard bar and plate sizes, common alloy grades. Ask your supplier what they hold in stock — stock material has no procurement lead time and no minimum order.
What it costs: a small design adjustment, and possibly a slightly thicker part than strictly needed.
Finishing is the line most often omitted from the comparison and most often the surprise at invoice. On small brackets, masking labour for powder coating can exceed the cost of the part itself.
What to do: specify finish only on the surfaces that are seen or that function. Avoid masking by designing around it — a hole that does not need to be precise can be coated through; one that does can be moved to a masked face. Choose as-machined plus deburr where appearance is not critical.
What it costs: appearance, on parts where nobody looks.
Setup is amortised, material is not. At 5 pieces, setup can be 50% of the price; at 500, it is 3%. But larger batches lock in inventory and reduce flexibility.
What to do: ask for price breaks at 1, 10, 100 and 1,000 in one RFQ and look at the curve. If the 100-piece price is less than 60% of the 10-piece price, the setup dominates and a larger first order is worth considering. If the curve is flat, the price is material and cycle time, and quantity will not help.
What it costs: working capital and the risk of committing to a design that changes.
| Change | Typical impact on unit price | Risk |
|---|---|---|
| General tolerance ±0.005 → ±0.1 mm, keep 3 critical dims tight | −15 to −30% | Low, if function is analysed |
| 4 setups → 2 setups | −20 to −35% | Medium, may need redesign |
| Deep pocket → 3× diameter depth, larger corner radii | −10 to −25% | Low |
| Titanium → 6061-T6 where loads allow | −60 to −80% | High, needs engineering sign-off |
| Full powder coat → as-machined + deburr on hidden faces | −8 to −15% | Low |
| 10 pieces → 100 pieces | −30 to −50% per unit | Low, inventory risk only |
Stacking the first, third and fifth changes on a typical bracket commonly produces a 40–50% reduction with no functional compromise at all — before any conversation about rate.
Sometimes. Forging or casting to near net shape reduces material removal and cycle time, but adds tooling and lead time. It becomes economic above a few hundred pieces, or where the material is expensive enough that scrap dominates.
Up to the point where setup is fully amortised. Beyond that, the curve flattens to material plus cycle time, and further quantity gains are small.
Rarely worth it. You lose the supplier's buying power, take on the risk of a material defect, and complicate traceability. It occasionally makes sense for exotic alloys or customer-supplied certified stock.
For machined parts, usually just the programming and setup again — often a few hundred dollars. For moulded parts, a tool modification, measured in weeks. The asymmetry is why design freezes matter more for tooling than for machining.
Tolerance. In our experience, the majority of RFQs arriving with a ±0.005 mm or ±0.01 mm general block have fewer than five dimensions that actually need it.
Once the specification is optimised, there is usually a real but modest margin left in the rate. How you go after it determines whether the supplier treats your next RFQ as a priority or as a chore.
"Can you walk me through what drives the price on this part?" gets a more useful answer than "can you do better". A supplier who explains that 40% is material and 35% is cycle time has told you where to focus — and has demonstrated that they have a model rather than a markup.
Suppliers have levers that cost them little: longer lead time, larger batch size, consolidated shipments, flexible delivery windows, payment terms. Offering one of those in exchange for a lower unit price is usually more effective than asking for a discount, because it improves their scheduling rather than their margin.
A quote at 500 pieces based on a verbal "we expect to grow" is priced as a 500-piece order. A purchase order for 500 with a scheduled release plan is a different conversation, and often a different number.
Sharing a competing number is legitimate if it is real and comparable. Using an unverified number to force a price down invites the supplier to match the number by removing a step — usually inspection — which is a saving you did not want.
This is unglamorous and consistently undervalued. A customer who pays on schedule is materially more attractive than one who pays late, and that shows up in how quotes are built, especially at the margin.
Look at tolerance first, then setups, then geometry, then material, then finishing, then quantity. The first two usually deliver more than any price negotiation, and neither requires changing suppliers.
We quote every RFQ twice — as drawn, and with the changes we would make — so you can see what the specification is actually costing. See our machining capabilities or send a drawing and ask for both numbers.
