The quality system stops defects. It stops learning too.

Here is an uncomfortable observation: the entire tariff argument over Chinese electric vehicles is people negotiating the entry fee while the race runs on a different track. I have lived the difference.

At a European tier-1 I once had a defect contained by lunchtime: suspect stock locked in the quarantine cage, a 100% interim check at end of line, supplier notified by the afternoon – all inside 24 hours. The approved control-plan change that made the fix permanent took six weeks. Three signatures. A customer notification. A re-run capability study. A plant in Shanghai can ship fix and paperwork inside that same window – not because its steel is cheaper, because its loop is shorter.

The variable that will decide this industry is not labour cost per hour. It is time from defect to countermeasure. Almost nobody in the tariff debate is talking about it.

What the system was built to prevent

IATF 16949, PPAP, VDA 6.3 – none of them are stupid. They encode a rational distrust of uncontrolled change, written for seven-year product cycles and tooling amortised over a decade. Uncontrolled change is exactly how bad parts reach a customer, so every change gets traced, evidenced and signed. I have sat on both sides of that freeze: drafting deviation requests early in my career, auditing the paper trails under VDA 6.3 later. The system does what it was built to do. Traceability holds. Chaos stays out.

The hidden tax is learn-cycle time. A PPAP submission freezes a snapshot of the process; by the time the approved snapshot comes back, the tooling and the volumes have moved. I have watched a "temporary" deviation quietly become permanent because the formal change behind it was still queuing for signature number two. One month of that wait burned €40,000 in sorting and premium freight. To protect a document, not a customer.

Where the Chinese model actually wins

This week's commentary on Chinese EVs out-selling and out-manoeuvring the German brands circles price, subsidies and wages. It misses the mechanism. Their advantage is not cheap steel, and it is not the absence of quality – they take recalls and field failures like everyone else. It is structural: vertical integration from battery cell to body shop, engineers with change authority standing at the line, and a habit of shipping small, frequent, reversible changes instead of quarterly lockdowns. Their quality loops close in days.

A European plant closes the same loops at documentation speed. When your competitor's correction cycle is 72 hours and yours is six weeks, you are being out-learned by an order of magnitude. No duty rate fixes that arithmetic.

Paper should be the receipt of learning, never the permit for it.

The fix you control this quarter

You do not fix this by abandoning PPAP. You fix it by decoupling the authority to correct from the paperwork that records the correction. Three moves, all runnable now.

• QRQC with teeth. 48-hour countermeasure authority at team-leader level, inside a defined sandbox – inspection frequency, tooling offsets, work sequence, containment. The person who sees the defect acts in the same shift; the QRQC sheet follows within 24 hours.
• A3 as receipt, not permission. Write the A3 once the countermeasure stabilises, as the record of what was learned. At WITTE Automotive the Q-Wall showed the fix days before the paperwork caught up, and failure costs fell because the loop closed – not because the folder grew.
• A living PFMEA. Reviewed at the line weekly, revisions controlled lightly. A PFMEA frozen at PPAP submission describes a process that no longer exists.

I tested the audit side of this at scale. At SNOP I built the QA/QC department of a 900-plus-employee greenfield plant from an empty office and deliberately kept the evidence structure lean – one clear page where legacy plants file five. It passed IATF audit without legacy bureaucracy, because auditors follow logic and evidence, not page count. The 70% defect-cost reduction and 98% customer satisfaction we achieved there came from loop speed: defect detected in hour one, countermeasure live in 24, standard updated in 48. Document volume had nothing to do with it.

The learn-cycle tax is not automotive-specific. Through FOREAST I have stood in heavy-industry plants – steel, optics, electronics – where a countermeasure everyone agreed on at the morning meeting waited two weeks for a signature chain. Same tax. Different sector.

The KPI that decides the next decade

Tariffs buy quarters, not decades. They put a price on the symptom while the disease – slow learning – compounds in the background. The headline quality KPI for the next ten years should be time-from-defect-to-countermeasure: measured in hours, posted at the line, reviewed with the same discipline as delivery. It is the one competitive variable a quality function controls this quarter, without capital approval and without waiting for Brussels, Berlin or Beijing to settle their arguments.

The quality system I have served for two decades was built to stop defects, and it does. Nobody designed it to stop learning – that came free, an unintended subscription fee. Strip the fee, keep the system. Decouple correction from documentation, measure the interval in hours, and the argument about Chinese electric vehicles stops being about tariffs and starts being about something a quality director can win. Somewhere tonight a change request is still waiting for its third signature.