Strategic sourcing is usually written as a category plan and then measured on purchase price. For industrial, defence, telecom and medical builds that run for years, that measurement is too narrow. The cost that hurts is not the price of the part, it is the build that stops, the qualification that has to be repeated, and the last time buy that was missed because nobody owned the part status.
A plan that is judged only on unit price will always favour the line that is easy to negotiate and leave the line that is hard to supply exactly where it was, which is usually the line that decides whether the program ships.
Where strategic sourcing goes wrong
Segmentation by spend is the first mistake. Spend concentrates on a handful of processors, converters and power stages, while the lines that stop a build sit in low spend categories: a connector, a custom magnetic, a filter on allocation. A category plan built on spend treats the tail as noise, and the tail is where the risk is.
A strategy copied from a template is the second. The right action for a leverage category, where several makers can supply and price moves, is not the right action for a bottleneck category, where one qualified source sets the date. Applying the same negotiation to both produces a price reduction on the easy lines and a single point of failure on the hard ones.
Qualification cost is the third and is rarely in the model. A second source that costs a month of engineering and a test lot is cheap on a ten year program and expensive on a two year one. Without that number the plan cannot decide where dual sourcing actually pays.
A second source named in a spreadsheet is the fourth. Named is not qualified: the alternative has to be tested against the same limits before it counts as a route. In an audit, a list of names with no qualification record is a list of intentions.
Contract terms are the fifth. Price validity, allocation in a shortage and the notice period for a part status change belong in the agreement. A price fixed for twelve months without an allocation clause returns the problem to the buyer at the worst possible moment.
Savings measured at the purchase order are the sixth. A ten percent reduction on a part that then needs expediting, re-inspection or a re-qualified alternative is not a saving. The number that matters is cost per good board shipped, which includes the scrap from partial kits and the labour of chasing dates.
End of life ownership is the seventh. Almost every plan has a status watch on paper, and few have a name against it and a date in the calendar when the list is reviewed.
What decides a strategic sourcing plan
| Decision | What it controls | What a weak plan looks like |
|---|---|---|
| Category segmentation | Which action fits which line | One negotiation for every category |
| Spend against risk | Where the effort goes | Spend reviewed, risk not |
| Qualification cost per category | Whether dual sourcing pays | Cost left out of the model |
| Second source state, named or qualified | Whether a route exists under pressure | Names on a list |
| Part status owner | Whether a last time buy is caught | Watch with no name against it |
| Price validity and allocation terms | What holds in a shortage | Twelve month price, no allocation clause |
| Tail management | How much of the BOM is unmanaged | Tail never reviewed |
| Kit completeness | Whether a build can start | Lines short at kitting |
| Savings basis | Whether the number is real | Purchase price only |
| Record and audit trail | Whether the plan survives an audit | Reports per vendor, not per lot |
| Last time buy and bridge stock | The cost of the final order | Discovered after the window closed |
How SUPERB PCBA supports strategic sourcing
Send the BOM with annual volume, program life and the lines that have already caused a stop. The reply segments the BOM by risk rather than by price, marks the lines where one qualified maker sets the date, states which alternatives are qualified and which are only named, and puts a date against each line. The operating model this fits into is set out in OEM procurement.
For the tail, the answer is a supply plan rather than a discount: bridge stock where the status is closing, a last time buy quantity sized on program life, and an approved alternative with the check that approved it. How that plan is kept alive is covered in component obsolescence management, and the change of shape when a supply base is reduced is covered in vendor consolidation.
The build side carries the same standards: IPC-A-610 Class 2 and Class 3, ISO 9001, ISO 13485 for medical programs, placement down to 01005, 3D SPI and 3D AOI, X-ray in 2D and CT with voiding below 5 percent, ICT and functional test, boards to 68 layers with impedance held to plus or minus 5 percent, and 730 days warranty on verified lots. Build is to print: the customer supplies the Gerber, BOM and assembly drawing, and SUPERB Assembly builds, inspects and tests to that package. Export runs under DAP, DDP or FOB.
The record a sourcing plan has to leave behind
A sourcing plan is only as good as the record behind each line: purchase order, lot number, certificate of conformance, incoming inspection result and the test data of the boards built from that lot. Parts are counted by X-ray on receipt, markings and date codes compared with maker data, decapsulation used where the risk justifies it, and aged stock checked for solderability. A substitution is made only with written approval from the customer, and SUPERB Tech keeps the mapping so a field return traces to a part and a lot.
Strategic sourcing questions
How often should a category plan be reviewed?
On quote and on status change, not on a calendar. A yearly review misses the part that closes its window in March, which is the case the plan exists for.
Should every line have a second source?
No. Lines with a long lead time, one qualified maker, or a qualification cost that blocks a fast change need one. Commodity passives with several equivalent makers rarely do, and duplicating them adds administration rather than cover.
How is a second source proved?
Tested against the same limits, on the same board, with the difference stated in the parameters that matter. A datasheet comparison is a shortlist, not a qualification.
What belongs in the supply agreement?
Price validity, allocation in a shortage, the notice period for a status change, and the basis on which the last time buy quantity is calculated. Without those four the agreement covers delivery, not supply.
Next step
Send the BOM with volumes, program life and the lines that have stopped a build. You get the BOM segmented by risk, the dual source map with its qualification state, and a last time buy quantity sized against program life rather than against a guessed run rate.