The sourcing moves. The capability doesn't.

Here is an uncomfortable observation from last week. Hyundai's India chief stood in front of the press and declared that India and South Korea offer a competitive alternative to China for EV supply chains. By Friday, that sentence had been pasted into sourcing decks from Stuttgart to Nagoya. Not one PPAP file changed anywhere on the planet. Strategy travels at news speed. Capability travels at run-at-rate speed, and the gap between those two speeds is where companies either earn their margin or donate it.

The quieter point came from the supplier side of the same news cycle: carmakers need quality parts, EV perks or not. Right, and the timing is no accident. Incentive money is pulling suppliers into the EV chain whose processes have never been proven at volume. The industry is redrawing the sourcing map faster than it is re-earning it.

The certificate crosses the border. The process doesn't.

A story from my FOREAST consulting years, which I keep telling because the industry keeps re-enacting it. A steel-sector client – the ArcelorMittal end of the market – had shifted production of a torque-critical component from a long-standing Asian plant to a new site an ocean closer to the customer. The VDA 6.3 audit file was immaculate: control plans current, work instructions signed, training matrices complete, calibration certificates aligned in rows. Then I asked to watch the joint get torqued on the actual line.

The live trace disagreed. The capability study in the file, a Cpk of 1.67, had been run on the spindle in the old plant. The new line had a different tool, a different clutch, and operators who had trained on the documentation rather than against a known-good curve. The trace wandered across the tolerance band like it was window-shopping.

The system was certified. The process was not.

A certificate is a photograph of the system. Capability is a property of the address.

Certification audits the documented system, and documentation is portable – it travels as a PDF. Capability lives in machines, moulds, fixtures and hands, and most of those stayed behind when the part number moved. An audit file full of studies run on equipment the new site has never owned is a history book, not evidence.

Relocation re-runs every failure mode

The second mistake is assuming the failure-mode library transfers. It doesn't. Every 8D in the archive was solved against a specific machine, tool, ambient condition and crew. Relocate the part and every one of those failure modes re-opens under new special causes – the humidity, the power quality, the sub-tier steel and the shift pattern all change underneath it. The old 8Ds are a starting hypothesis, not immunity.

So every transferred part number gets the full treatment: a rebuilt PFMEA, because the process and its causes both changed; twin samples; re-PPAP; capability studies from zero on the new line. A re-PPAP on a moderately complex component – full layout, capacity studies, run-at-rate – runs €20,000–€50,000, and a safety-critical part with full validation clears €100,000. At WITTE Automotive we lived the counter-arithmetic: the QRQC and A3 discipline that cut our failure costs so hard worked because the money went into detection before volume, not after escape. One escaped lot – sorting, containment, premium freight, penalties – burns €250,000, and it never arrives alone.

I learned the deeper version of this at SNOP, building a greenfield QA/QC department for a plant of 900-plus people. A new site's biggest risk is what it does not yet know about itself: no 8D library, no operator memory, nobody who can say "that press sounds wrong today". So we industrialised learning instead of assuming it – daily QRQC at the line, a physical problem wall, capability data on every process before volume release.

A relocated supply region is a new site with a flag. It does not inherit immunity. It grows it, part number by part number.

Qualify the region like a part

If geography is now a strategic input, qualify it like a characteristic – phase-gated, with exit criteria, not a press release. Three moves separate the serious from the decorative:

• Resident quality engineer for the first 90 days. Not a visit – a desk at end-of-line, every shift start.
• Three-lot containment on critical characteristics. One hundred per cent inspection for the first three production lots, stepping down on data, not on comfort.
• Shared 8D rights. The customer can open and drive a corrective action in the supplier's system without negotiating whose logo goes on the form.

And the escalation ladder gets signed before the first container ships: named people, response clocks, a level that reaches both executive teams inside 24 hours. None of this is bureaucracy. Leading quality initiatives across a 2,000-plus workforce over multiple sites taught me that ladders built before volume hold, and ladders improvised after the first escape are archaeology – you dig through them to work out who knew what, and when. Behind all of it sits the incentive question. Production-linked schemes subsidise capacity, not competence; the money buys parts and has never bought capability. Budget the qualification or pay the failure cost. Those are the only two line items on offer.

None of this argues against India or South Korea – both countries host plants that would embarrass half of Europe. It argues against confusing a strategy statement with a qualification, which is what both threads of last week's news were about: the podium declaration on one side, the supplier asking for quality money instead of perks on the other. Sourcing is a quality decision wearing a procurement coat. The map changed; the physics didn't. China+1 will be won by the companies whose quality engineers sat at the table before the contract was signed, not after the first escape proved the press release optimistic.