Why magnets resist ordinary purchasing
Most commodity purchasing assumes short, stable lead times and a liquid spot market. Permanent magnets break all three assumptions at once, and the mismatch is what produces the familiar cycle of emergency air freight followed by excess inventory.
The cost that never appears in the price comparison
A spot-buying policy on a long-lead part carries an implicit cost: the expedites, the line stoppages, the substitutions accepted under pressure, and the price paid for whatever was available rather than what was specified. That cost is real and it is usually larger than the inventory carrying cost of the program that would have prevented it — but it lands in operations and engineering budgets rather than in the purchase price variance report, so it stays invisible.
The six supply structures
These are not competing philosophies; they are different points on a curve trading cash, risk and price. Most manufacturers use two or three simultaneously across different part profiles.
| Structure | Stock held by | Title held by | Cash impact on you | Typical price position |
|---|---|---|---|---|
| Spot purchase | Nobody | — | None until ordered | Highest — no volume leverage |
| Blanket PO, scheduled releases | Supplier | Supplier until release | Low — pay on delivery | Volume pricing on committed quantity |
| Min/max bridge stock | Supplier | Supplier | Low | Volume pricing plus a stocking element |
| Consignment | You, on your site | Supplier until consumed | None until consumed | Slight premium for carrying cost |
| Vendor-managed inventory | You or supplier | Varies by agreement | Depends on title terms | Priced against the replenishment service |
| Buy-and-hold | You | You | Full, at purchase | Lowest unit price, highest cash and obsolescence risk |
Blanket purchase orders in practice
The workhorse structure, and the one most often written badly. A blanket order commits to a total quantity over a period at an agreed price, with delivery called off in scheduled releases. It gets you volume pricing without volume inventory — provided the terms are specific.
The six terms that decide whether it works
The clause buyers forget
Liability on cancelled schedule. A blanket order authorises the supplier to buy material and start production ahead of your releases — that is precisely how it shortens your effective lead time. If you then cancel, someone owns that material. Define the exposure explicitly: liability limited to finished goods, work in process, and raw material committed within the agreed production horizon. Without it you are either exposed to the full commitment or you have a supplier who will not build ahead, in which case the blanket order is delivering nothing.
Consignment and vendor-managed inventory
Both move inventory closer to the point of use. They differ in who owns it and who decides when to replenish, and the two questions are independent.
Consignment
Supplier-owned stock physically located at your facility. You draw from it; title transfers at consumption; you are invoiced for what you used. Physical availability is immediate, and the inventory does not sit on your balance sheet.
Vendor-managed inventory
The supplier takes responsibility for maintaining agreed stock levels, replenishing against visibility of your consumption rather than against your purchase orders. VMI can operate with either party holding title; the defining feature is that the replenishment decision moves to the supplier.
- Requires real consumption data. VMI on a monthly emailed spreadsheet is not VMI — it is a slow blanket order. Give the supplier usable visibility or do not adopt the model.
- Min/max levels are the contract. Agree them per part number, review them quarterly against actual demand, and define what happens when the supplier misses a minimum.
- Works best on stable, repeating demand. Highly variable or campaign-based consumption is better served by bridge stock against a blanket order.
- Accountability must be explicit. If the supplier owns replenishment, the supplier owns the stockout — the agreement should say so and state the remedy.
Bonded and bridge stock
A lighter-weight middle option: the supplier holds an agreed quantity of your specific part, finished and inspected, in domestic warehousing, available on short call-off. You commit to consume it within a defined window. There is no consignment accounting and no on-site space required, but the lead time on those units collapses from months to days — and for imported parts, that stock has already cleared customs and any export licensing, removing the border risk entirely. Radial Magnets operates this structure for key accounts.
Sizing the buffer
Whatever the structure, someone has to decide how much. Sizing by intuition produces both stockouts and dead stock, usually on the same shelf.
Safety stock with variable lead time
The simple formula assumes the lead time is fixed. For magnets it is not, and ignoring that is why buffers sized on the textbook version still fail. Use the combined form:
combined variability
safety stock = Z × √[ (LT × σD²) + (D² × σLT²) ]
where Z is the service factor (1.65 for 95%, 2.05 for 98%, 2.33 for 99%), LT is average lead time in weeks, σD is the standard deviation of weekly demand, D is average weekly demand, and σLT is the standard deviation of lead time in weeks.
worked example
A magnet consuming 2,000 pieces per week (σ 400), on a twelve-week average lead time that has varied by roughly three weeks either side, at a 98% service target:
- Demand variability term: 12 × 400² = 1,920,000
- Lead time variability term: 2,000² × 3² = 36,000,000
- Combined: √(1,920,000 + 36,000,000) ≈ 6,160
- Safety stock: 2.05 × 6,160 ≈ 12,600 pieces — about six weeks of cover
Note which term dominates
In that example the lead time variability term is roughly nineteen times larger than the demand variability term. The buffer exists almost entirely because delivery dates move, not because usage moves. This is the mathematical case for supplier reliability being worth more than unit price: reducing σLT from three weeks to one would cut the required safety stock by more than half, and that saving is permanent, whereas a price concession is annual.
Carrying cost, honestly stated
Annual carrying cost is typically 18–28% of inventory value once capital, warehousing, insurance, handling, obsolescence and shrinkage are counted. On magnets, obsolescence risk is lower than on electronics — a correctly stored magnet does not go stale — but it is not zero, because design changes strand tooled parts that have no alternative use. Use a real figure for your business rather than the default 25%, because it determines which structure is actually cheapest.
Commercial terms that decide the outcome
Price is the term buyers negotiate hardest and the one that most often turns out not to have been the important one.
Price firmness versus indexation
| Approach | How it works | Suits | Risk |
|---|---|---|---|
| Firm fixed for the term | One price for the full commitment period, typically six to twelve months | Budget certainty; programs with fixed sell prices | Supplier prices in a risk premium; you may sit above market if inputs fall |
| Indexed to rare earth pricing | Base price adjusts periodically against a published NdPr or Dy oxide index, usually with a dead band and a cap | Long agreements; large volumes where a fixed-price premium is expensive | Requires an agreed index and mechanism; exposes you to upside moves |
| Fixed with a reopener | Firm unless the index moves beyond a threshold, at which point both parties renegotiate | Most manufacturers — the practical middle | Reopener thresholds must be symmetric, or it is a one-way ratchet |
Make the escalation clause symmetric
A clause allowing price increases when rare earth indices rise, with no corresponding decrease when they fall, is common and is worth pushing back on. If the mechanism is genuinely about passing through input costs, it passes through in both directions. Insist on the same threshold, the same index and the same review cadence for movements in either direction — and specify which published index, since they do not track identically.
The rest of the term sheet
- Tooling ownership and location. If you paid for it, the agreement should say you own it, name the facility holding it, and give you the right to have it transferred.
- Change control. No change to material source, coating process, mill or magnetizing method without written notification and, on qualified parts, requalification. This clause is what stops a silent process change from becoming a field failure.
- Second source rights. The right to dual-qualify without penalty, and access to the drawing package and tooling required to do it.
- Documentation obligations. Which certificates ship with every lot, specified in the agreement rather than requested per shipment — see certificates and compliance documents.
- Quality remedies. Response times for nonconformance, containment obligations, and who pays sorting costs — see incoming inspection.
- Capacity commitment. On critical parts, an agreed capacity reservation is worth more than a price break.
Second sourcing, done properly
Dual-qualifying two distributors who both buy from the same mill provides no supply resilience whatsoever — it duplicates commercial overhead while leaving the single point of failure untouched. A meaningful second source is a different manufacturing facility, ideally in a different country, with its own tooling, its own first article and its own qualification record. Ask which mill produces the parts, and verify it against the material certificates you receive.
Storing magnet inventory
An inventory program only works if the stock is still good when you draw on it. Magnets do not have a shelf life in the ordinary sense — a correctly stored magnet is unchanged after years — but the failure modes that do exist are entirely preventable.
| Risk | Mechanism | Control |
|---|---|---|
| Coating corrosion | Humidity attacks any coating breach; NdFeB then corrodes from the grain boundaries outward | Controlled humidity, sealed packaging retained until issue, original desiccant kept in place |
| Chipping and cracking | Magnetized parts snap together in transit and handling; sintered material is brittle | Interleaving spacers, foam separators, original packaging, no bulk decanting |
| Self-demagnetization | Low-coercivity materials lose flux without a closed flux path | Keeper plates on alnico; not required for NdFeB, SmCo or ferrite |
| Field exposure to other stock | Attraction across shelving damages adjacent inventory and injures handlers | Segregated storage area, ferrous-free racking nearby, signage per site standards |
| Lot identity loss | Mixed lots break certificate traceability and destroy any failure investigation | Lot-controlled locations, no commingling, FIFO enforced by location rather than by intent |
| Handling injury | Large magnets attract with crushing force | Gloves, eye protection, controlled separation technique, documented handling procedure |
Lot control is not optional on a program part
The moment stock from two lots is commingled in a bin, the link between your parts and their material certificates is gone. If a magnetic or dimensional problem surfaces later, containment expands from one lot to everything in the bin and everything built from it. On qualified and safety-relevant parts, lot segregation is what keeps a quality event small.
Choosing the right structure
| Part profile | Recommended structure | Why |
|---|---|---|
| Low volume, catalogue size, non-critical | Spot purchase | Available from stock in days; a program adds overhead for no benefit |
| Steady volume, custom tooled, single source | Blanket PO + bridge stock | Volume pricing plus a buffer against the lead time that cannot be compressed |
| High volume, stable demand, floor-level consumption | VMI or consignment | Removes purchasing transactions entirely and keeps cash off the balance sheet |
| HRE-bearing grade, export-licence exposed | Domestic bridge stock, extended cover | Buffer sized against regulatory interruption, held already cleared |
| Automotive or medical qualified part | Blanket PO with change control and capacity reservation | Requalification cost makes supply continuity worth more than unit price |
| Program approaching end of life | Last-time-buy against a firm requirement | Avoids stranded tooling costs and a rushed final order at a poor price |
| Campaign or project demand, lumpy | Blanket PO, long release notice | Commitment gets pricing; long notice keeps the supplier able to plan |
What to bring to the conversation
A supplier can only structure a program against real data. Come with:
- Twelve to twenty-four months of usage history by part number, monthly — actual consumption, not purchase orders.
- A forward forecast with an honest confidence level attached. An overstated forecast produces a commitment you will pay for.
- Criticality classification. Which parts stop a line and which do not. Buffers should not be uniform.
- Your carrying cost percentage and cash constraints, so consignment versus buy-and-hold can be evaluated honestly.
- Known design changes or program end dates inside the commitment window.
- Current lead time performance from existing suppliers — the variability figure that sizes the buffer.