Nine hundred and fifty-five thousand vehicles. I ran the arithmetic out of habit, the way you check a scrap report before your first coffee: at a conservative €200 per unit across logistics, dealer time and administration, Stellantis is holding a nine-figure invoice before anyone has turned a wrench. The uncomfortable observation came with it. Nobody in that chain lacked data. The complaint log had entries. The warranty returns had a shape. Somebody's line-reject bin had the parts. A recall is the only customer complaint that arrives with its own press release – by the time it makes headlines, the organisation has already been shown the same story in three cheaper formats and declined to finish reading it.
The rule of ten never suspends itself
The rule of ten is the oldest arithmetic in quality, and it does not pause for quarter-end. The same defect costs cents at the station, tens of euros at end-of-line once rework labour and freight attach, hundreds at the dealer, four figures per unit once it matures into a field action with legal overhead on top. Plants budget accordingly – cents for detection, millions for escape – because detection is a line item somebody must defend in the cost workshop, while escape hides in contingency and goodwill. That inversion is the disease.
In two decades of plant audits I have learned to look past the PPM chart. What predicts trouble is speed: how fast a suspicious part travels from an operator's glove to a containment decision. A nine-figure recall is the rule of ten arriving with interest.
Your early-warning system already exists
You do not need to buy an early-warning system. You are already running three: the reject bin, the complaint log and the warranty return pile. Each is a leading indicator the field will eventually confirm at retail price. The failure is rarely collection; it is interpretation. Complaint logs fill with soft language – customer reports intermittent noise – and get triaged by people measured on closure rate rather than pattern recognition. Warranty returns tagged NTF, no trouble found, are treated as noise when they are often the loudest signal in the building: the failure mode exists, your test bench simply does not reproduce it. Mining those logs for clusters is now trivially cheap. The barrier is not technical. It is that nobody is paid to be frightened by text.
Then there is the word closed. I have read hundreds of 8D reports, and closed is the most load-bearing word in the document. Too often it means the paperwork is complete: containment shipped, corrective action marked implemented, verification performed by the absence of further complaints. The failure mode was never killed; it was documented until it stopped being inconvenient. A KPI on closure rate gets you exactly that – closures, not corrections.
Every recall begins life as a nonconformance nobody was embarrassed by.
QRQC is escalation with teeth
QRQC – Quick Response Quality Control – gets described as a tool, which undersells it. It is escalation with teeth: a nonconformance triggers analysis on the spot, with the parts, the data and the people who touched them, and management is standing at the board within the hour. Not briefed by Friday. At the board, within the hour, looking at the actual failed component. Containment gets measured in hours, not quarters, because the only clock that matters starts at detection and stops at containment.
I built a greenfield plant's quality organisation around that clock – nine hundred plus employees, QA and QC from bare concrete upwards. QRQC up front, A3 discipline behind it, and the number we watched was not reports closed but detection-to-containment time, published daily where everyone could see it shrink. Defect costs fell seventy per cent. We went a full quarter without a single critical customer escalation – not because defects stopped appearing, but because they died young, at the cheap end of the rule of ten. The same discipline had already worked at Witte Automotive, where QRQC, A3 and a Q-Wall that made every escape visible in real time drove our failure-cost reduction. Across two decades in automotive and aerospace I have not seen another lever that pays like it.
The cultural shift is the real deliverable. A plant that documents nonconformances complies; a plant that is embarrassed by them improves. A part sitting on the QRQC board for three days is not a process delay – it is management credibility leaking at a visible rate. And once operators watch the quality director answer for a rusty bracket before lunch, escalation stops being a career risk and becomes the fastest route to being left alone.
The field reports eventually
None of this is new; Toyota built most of it decades ago. What changed this week is the scale of the invoice. The field is the most honest reporter in your value chain and the most expensive: it never misses a deadline, it sets the price by how long you made it wait. Stellantis will now spend nine figures discovering what its own complaint log knew months ago. Every plant manager reading that headline owns an identical loop – running slower than it should, priced by the same tariff.
Key takeaways
- Measure detection-to-containment in hours and publish it daily at the QRQC board – it is the one metric that predicts recall exposure.
- Audit "closed" 8Ds monthly for failure-mode evidence, not paperwork completeness; reopen anything verified only by silence.
- Treat NTF warranty returns and line-reject bins as leading indicators, reviewed with the same cadence as customer PPM.
- Run your own recall arithmetic now: units in the field multiplied by cost per unit. That number is the honest budget for your detection spend.
A recall is not a bolt from the blue. It is a feedback signal that waited in queue until it could hire a press officer. Compress the loop this quarter – hours from the glove to the containment decision, not weeks from the complaint form to the corrective-action counter – and the recall your successor would have managed never happens. The field reports eventually. The only decision you actually get is the price.