Here is an uncomfortable observation from someone who has stood up quality in a greenfield plant rather than merely audited one: when a €130K flagship throws a defect, the market forgives a sensor ghost long before it forgives the brake pedal. Maextro, the Huawei–JAC marque, is answering for exactly that – a reported brake-pedal failure on a car priced against the S-Class.

No algorithmic ambiguity here. No corner case buried in ten million lines of code. The oldest, most understood failure mode in the book escaped from one of the newest plants on the market. That is not bad luck. It is a launch-quality decision someone made, almost certainly with a spreadsheet open.

Launch cars are not production cars

I built the QA/QC department of a greenfield plant for 900-plus people from an empty floor upward, and the most useful thing I carried into that ramp-up was institutional distrust. A launch-year plant is a different factory wearing the same logo. Suppliers arrive at T1 maturity, operators are in week three of training, torque tools have not been characterised against real joint behaviour, and the gauges are fresh off their calibration certificates with no stability history behind them. The first ninety days of a new plant run at the worst capability that plant will ever record – precisely when the price point, and therefore the exposure, sits at its highest.

The industry's answer is old and unglamorous: enhanced containment. GP-12 in GM dialect, safe launch in everyone else's. 100% inspection on defined characteristics, documented exit criteria, release by data.

The exit is the part people cut. Containment ends when capability is demonstrated, never promised – clean months, capability indices on the characteristic, verified by someone whose bonus does not depend on the answer. Not when volume pressure arrives. Not when the plant is declared mature. At that greenfield site we held launch containment well past the point fashion would dictate, and we recorded zero critical customer escalations in the first quarter. The containment was never the insult. The escape would have been.

A premium badge does not lower the inspection burden. It raises the price of every escape.

The safety-critical single point changes the control plan

Severity 10 in a PFMEA means the failure occurs, the driver cannot compensate, and people get hurt. A pedal box is a single-point part. No redundant channel behind it, no limp-home mode that survives a pedal which no longer transmits force. Across two decades of IATF 16949 and AS9100 work I have applied one rule without exception: when severity is 10 and redundancy is zero, detection stops being a layer of defence and becomes the entire defence.

That changes the control plan in specific ways. Error-proofing on the affected station is verified every shift, not every month – a red-rabbit challenge at shift start, logged, with a defined reaction plan for the morning it fails to trip. A poka-yoke unchallenged this morning is a poka-yoke that does not exist. Critical characteristics get 100% verification with automated records instead of sampling, and traceability has to survive down to the serial number, because on a €130K car the recall-scope decision gets made from that data at three in the morning.

The first 72 hours decide the brand

When the call comes, the clock is the strategy. QRQC discipline, nothing fancier: the physical part on the table within the hour, is/is-not analysis at the shop floor, contain before you diagnose, verify before you ship. The stock wall goes up within 24 hours – plant, finished-goods warehouse, in-transit pipeline, dealer stock. Every hour of delay converts a defect into a story. At Witte I ran Q-Wall containment alongside QRQC and A3, and the failure-cost reduction we delivered came less from the boards on the wall than from the fact that the clock started at detection, not at escalation.

Then there is the communication, which quality people chronically treat as public relations and therefore beneath them. It is not. Honest, specific, early communication with the customer is a de-escalation tool of the first order. Stellantis is watching three-month-old cars for proof that its fix holds while carrying a €4.1 billion bill for the older ones – the arithmetic of waiting is on public display. New brands survive escapes with speed and candour, never with silence. And in a market where the regulator has set a year-end safety deadline for the entire industry, silence does not just move slowly. It compounds.

Key takeaways

  • Treat the first ninety days of a new plant as a different product: 100% containment on safety characteristics, with data-based exit criteria – capability demonstrated, never promised.
  • Severity-10 single-point parts get error-proofing verified every shift. A poka-yoke without a red-rabbit challenge is decoration.
  • The stock wall goes up within 24 hours across plant, pipeline and dealer stock. Every hour of delay converts a defect into a narrative.
  • In the first 72 hours, candour is cheaper than lawyers and faster than regret. Speed and honesty are a young brand's only recoverable assets.

Charge whatever the market will bear – the process still starts at serial number one. A young plant does not get to skip maturity; it chooses whether to buy the time with containment or bill it to the customer. Enhanced inspection at launch is not an admission of weakness. It is an acknowledgement of what every ramp-up statistic already says. Skip it, and the customer runs the durability trial for you – at flagship prices, in public, with the regulator watching. That trial has one judge and no appeals.