The Real Cost of Holding Machined Parts in Inventory

Most purchase orders for machined parts are decided on two numbers: unit price and promised delivery date. Carrying cost almost never makes it onto the comparison sheet. For a part you buy once a year that omission is harmless. For a part you buy every quarter in the thousands, it can quietly exceed the entire gap between the cheapest and the most expensive quote you received.

If your company owns the stock — not consignment, not supplier-managed inventory — you are already paying this cost whether or not anyone booked it. What follows is the rate, the parts that behave differently, and the order quantity decision.

What carrying cost actually includes

Carrying cost is usually expressed as an annual percentage of inventory value. Published ranges cluster between 18% and 28% for manufactured goods, and the upper half of that range is normal for metal parts with long tooling lives and slow revision cycles.

Two components matter most. The first is obsolescence. A part that sits for eleven months and then gets a revision B is worth its scrap value, not its book value. If your engineering change rate on a commodity bracket is 8% a year, you are effectively writing off a slice of every large order you place. The second is handling damage. Every time a box is opened to pull twenty pieces, someone touches a machined surface, and anodized or bead-blasted cosmetic faces pick up marks that a bare aluminium face would survive.

Why machined parts are not a generic inventory class

Two properties make metal parts behave differently from, say, fasteners or packaging.

Value density is high. A 6061 bracket with 45 minutes of three-axis time in it might carry $22 of value in a package the size of a paperback, so you are storing money rather than volume and rent is rarely the dominant term.

Revision sensitivity is real but uneven. A housing tied to an industrial enclosure can stay unchanged for six years. A bracket inside a consumer product can change three times in eighteen months. Treating both with the same safety stock policy is how companies end up with $40,000 of unusable parts in a corner.

See also: Air vs Sea Freight Cost for Machined Parts: When Air Actually Wins and What Drives CNC Prototype Cost, Line by Line.

The order quantity math most buyers skip

The classic economic order quantity model balances ordering cost against carrying cost. In a machine shop context, "ordering cost" is not the cost of raising a PO — it is the setup, programming amortization and inspection cost of running a batch.

Here is a worked example. You buy 5,000 units a year of a $18 machined bracket. Your carrying rate is 24% a year. The shop charges a $340 setup and first-article cost each time you release a batch, and lot sizes above 2,500 get a 4% price break.

• Two orders of 2,500: average stock 1,250 units, value $22,500, carrying cost $5,400 a year, setup cost $680, total $6,080.
• Four orders of 1,250: average stock 625 units, value $11,250, carrying cost $2,700 a year, setup cost $1,360, total $4,060.
• One order of 5,000 with the 4% break: average stock 2,500 units, value $45,000 at $17.28, carrying cost $10,368 a year, setup cost $340, total $10,708.

The quarterly release wins by about $2,000 a year on a part worth $90,000 annually — a 2.2% swing on spend, achieved without negotiating anything. The 4% price break looks attractive in isolation and is worth roughly $3,600, but it costs $6,300 in carrying and setup, so it loses.

Where the money hides in practice

Carrying cost is not evenly distributed across a warehouse. In most plants we visit, three pockets hold most of it.

The "just in case" buffer. A buyer adds 15% to an order because a previous supplier missed a delivery in 2024. That buffer then sits untouched for two years and is reordered anyway because the ERP sees consumption. The rate is a symptom, not the disease; the cure is supplier delivery data, not a bigger safety stock.

The dead shelf. Parts for a cancelled program, a superseded revision or a decommissioned machine. Nobody has authority to scrap them, so they are counted every quarter at full value. A quarterly review with a named owner who can authorise write-off recovers most of this value through scrap sale.

Inbound inspection queues. Material that has arrived but has not cleared incoming inspection is inventory too, and it is the worst kind because it cannot be used. You are effectively holding five extra days of stock on every part number at once.

Changing the decision without changing the supplier

You do not need a new supplier. Four levers work inside an existing relationship.

1. Move to scheduled releases against a blanket order. You commit to annual volume, the shop holds raw material, and you call off monthly. Your average finished-goods inventory falls; the shop's raw material position improves because it buys mill stock once.
2. Ask for a kanban or two-bin arrangement on high runners. A small buffer at your line, replenished on consumption, removes the forecast error from the order quantity.
3. Price the revision risk explicitly. If a part is likely to change, order in smaller lots even at a higher unit price. The 4% premium is cheaper than an 8% obsolescence rate.
4. Share your real consumption data. Shops quote lead time and lot size partly on assumptions. A twelve-month usage history with the peaks visible lets a supplier plan tooling and material, which usually converts into a shorter lead time at the same price.

If you are rebuilding a make-or-buy or resourcing model, our capabilities overview explains what batch sizes and finishing options are practical at each volume band.

Frequently asked questions

Is 25% a reasonable carrying cost assumption for a mid-size manufacturer?

For metal parts with moderate revision risk, 20-26% a year is a defensible planning figure, and many finance teams use a single corporate rate around 22%. The right number depends on your cost of capital and your observed obsolescence rate. Start at 25% if you have never measured write-offs.

Does carrying cost apply if the supplier holds consignment stock at my plant?

Partly. You avoid the capital charge until you consume, but you still carry space, insurance and handling exposure, and consignment agreements often include a minimum take-or-pay commitment that transfers obsolescence risk back to you. Read the termination clause before treating consignment as free inventory.

How does this change for low-volume, high-mix production?

At low volume the setup cost dominates and larger lots are usually correct. The exception is when a part has a short engineering life or an expensive material; then the obsolescence term can exceed the setup saving. A useful rule of thumb is that any part with a revision probability above 20% a year should not be bought in lots larger than one quarter of demand.

Should I include carrying cost in a supplier comparison?

Yes, but state the assumption openly rather than hiding it in a score. Two suppliers quoting different lot sizes and lead times are effectively selling you different inventory positions, and comparing bare unit price rewards the supplier who pushes the most stock onto you. Show it as a separate line, and confirm with your finance team how inventory is valued before you build the number into a business case.

Related reading

• Material Cost in Machining: The Share Most Buyers Underestimate2026-10-23
• Expedite Fees in Manufacturing: What Rush Delivery Actually Costs2026-10-25
• Machining Cycle Time Calculation: A Worked Example on an Aluminum Bracket2026-10-22