Comparing Machining Suppliers on Total Cost of Ownership, Not Price

Every sourcing decision starts with a spreadsheet column called "unit price" and ends, too often, with a post-mortem six months later about air freight, sorting costs and a line that stopped for two days. A total cost of ownership supplier model exists to move that post-mortem forward, into the decision itself.

The mechanics are simple: list every cost the decision actually creates, normalise each to a per-unit figure, and weight the categories you care about. The difficulty is data. Most companies can get piece price in a day and cannot get their own inspection labor cost without three meetings, so the structure below is built to work with imperfect numbers.

Why piece price misleads

Piece price is quoted at the factory gate. Everything after that gate — freight, insurance, duty, clearance, inland transport — is invisible in the comparison but real in the P&L. On a $4.20 machined housing, landed cost can easily be $4.70.

Piece price does not include the cost of the supplier being wrong. Scrap found at your incoming inspection is your labor, your freight and your schedule. Scrap found at your customer is all of that plus a containment exercise and a credibility hit.

The line items that belong in the model

Here is a structure we use, with plausible values for a 10,000-piece annual requirement of a 38 mm aluminium housing. Two suppliers are compared: an overseas supplier at $4.20 and a domestic supplier at $5.90.

The gap has closed from 40% on price to 15% on total cost, and the domestic supplier still loses. That is a useful outcome: the model did not force a predetermined answer, it quantified the real premium for proximity. If your business case for local supply depends on a 40% price gap, you now know you are paying $0.92 per unit for shorter lead time, and you can decide whether that is worth it for this part.

Two adjustments often flip the result. Change the part to a $60 precision component with a 3% overseas defect rate uncovered at final assembly, and the scrap line moves from $0.08 to $1.80. Shorten the program to an 18-month build with a likely revision, and the expedite and obsolescence risk on a 12-week pipeline becomes the dominant term.

See also: Tooling Cost Amortization and Fixture Ownership and What Drives CNC Prototype Cost, Line by Line.

Setting the weights

Weights convert the line items into a single score, which is useful for ranking and dangerous if left unexamined. A workable starting set for machined parts:

• Piece price — 40%. Still the biggest single factor at stable volume.
• Quality — 20%. Scrap, returns, corrective actions, first-pass yield on new parts.
• Logistics and duty — 15%. Landed cost delta, transit variability, customs friction.
• Delivery performance — 10%. On-time percentage and, more importantly, the spread.
• Engineering responsiveness — 8%. Quote turnaround, DFM feedback quality, willingness to take a call at your end of the day.
• Compliance and administration — 7%. Documentation quality, declaration turnaround, audit cooperation.

Change the weights to match the part. A cosmetic front panel for a launched product should weight quality and delivery higher, because a defect stops a revenue line.

Score each supplier 1-5 per category, multiply by the weight, and sum, keeping the raw dollar figure alongside. If the score and the dollar total disagree, you have a weighting problem worth discussing, not a data problem to hide.

Getting the data without a six-month project

Most of the numbers above are already inside your business.

Freight comes from your forwarder's invoices, divided by units shipped. Duty comes from your customs broker; use the rate that actually applies to your HS classification and confirm it with them rather than with a search engine. Inspection hours come from your quality team's time records or, if you do not track them, from a two-week time study. Scrap comes from nonconformance records plus rework labor. Expedite cost comes from air freight invoices with a reason code attached.

The two hardest to get are line stoppage exposure and management time. For stoppage, take your internal cost per hour of a stopped line, multiply by the hours lost per incident and by incidents per year for that part family. For management time, ask the buyer and quality engineer to log hours per supplier for one month; the number is usually larger than anyone expects.

Two asymmetries are easy to miss. Quality escapes have a tail: a defect that reaches a customer generates containment and chargebacks that land a year later. And payment terms are financing — a supplier offering 60 days versus 30 is effectively funding 30 days of your inventory, worth about 0.75% of annual spend at a 9% cost of capital.

Running the comparison without breaking the relationship

Do not send the scorecard to suppliers. Use it internally, and translate the output into specific asks: a tighter on-time commitment, a different inspection regime, a buffer of finished goods, a documented corrective action process. Suppliers improve when they are told which line item is costing them the order, and almost never when handed a matrix with a 3.4 out of 5 next to their name.

Where the model changes the decision, document the assumption. If you choose the domestic supplier at a $1.60 premium per unit, write down what you are buying for that money — six weeks less pipeline, no customs exposure, a four-hour drive instead of a five-week ocean leg.

For the process side of this — how quotes are structured, what a shop includes or excludes, how lead time is committed — see how we work and our capabilities. Our own quotes run at 24 hours for most geometries, which keeps at least one variable stable in your comparison.

Frequently asked questions

How many suppliers should a TCO comparison cover?

Two or three is enough. Beyond three the marginal information gain is small and the data burden grows, because every supplier needs its own inspection, freight and nonconformance history. If you have no history with a new supplier, use your incumbent's defect and expedite rates as a proxy and flag the assumption in the model.

What if I cannot get accurate freight and duty numbers for a new supplier?

Use a bounded range rather than a point estimate. Get one indicative quote from your forwarder for the lane and weight, apply the duty rate your broker confirms for the HS code, and present the result as a range. A model with honest error bars is more useful than false precision to four decimal places.

Does TCO apply to low-volume or one-off orders?

Partly. At 50 pieces the fixed costs — freight minimums, brokerage fees, inspection setup — dominate, so the model reduces to landed cost plus the cost of a defect. Weight quality and engineering responsiveness much higher and drop the inventory-related terms, which are negligible on a single lot.

How often should the model be refreshed?

Annually at minimum, and immediately after any significant event: a tariff change, a freight spike, a quality escape, a shift in your own labor rates. The line items that move fastest are freight, duty and defect rate; piece price moves slowest, which is precisely why companies over-weight it. Confirm classification and duty treatment with your customs broker before acting on any number in the model.

Related reading

• Material Cost in Machining: The Share Most Buyers Underestimate2026-10-23
• Scrap Rate in Machining: Where Losses Come From and Who Pays2026-10-24
• Packaging Machined Parts: Protecting Precision Surfaces in Transit2026-10-27