The survey scores 90 days. The warranty pays for ten years.

Ford has said out loud what OEMs normally keep buried in a strategy deck: it intends to be number one in the J.D. Power Initial Quality Study. From the boardroom, that reads as ambition. From the shop floor, it reads as a weather warning. I spent two decades on the supplier side of this industry – door-latch and lock systems at WITTE Automotive, body-in-white at a SNOP greenfield plant – and I know what such a sentence becomes two management levels down. It stops being about building better cars and becomes about what gets counted, when it gets fixed, and whose scorecard absorbs the difference.

The mechanics matter. The study scores problems per 100 vehicles – PP100 – as reported by owners during their first ninety days with the vehicle. It is a perception survey: a failed fuel pump and a radio-preset hierarchy the owner cannot work out each count as one problem. The score has no engineering conscience and no memory. Everyone being measured on it knows both facts.

What a ninety-day window does to honest people

Nobody sets out to cheat a survey; pressure finds the cheapest path. When bonuses hang on a number counted over ninety days, three behaviours appear with metronome reliability.

• Classification games. Claims migrate from "functional failure" to "customer dissatisfaction" to "operates as designed" – each step cheaper than the last. "No fault found" becomes a growth category. Nothing about the part changes. The chart does.
• Fixes timed to the calendar. Running changes parked at the model-year break. Containment applied to the build weeks that map to surveyed vehicles, then quietly lifted. The remedy is real; its timing is a negotiation.
• Software-only remedies. The reset that substitutes for harness rework. The update queued for month four. The first ninety days stay clean; the root cause stays employed.

The deeper flaw is structural. The survey has amnesia by design. Corrosion creeping through a hem flange, a solder joint ageing under thermal cycling, a fatigue crack that arrives at 80,000 km – none of these exist at day ninety. Initial quality does not lie about them. It never asks.

A defect reclassified is a defect deferred – and the warranty bill pays the interest.

The cascade lands on your supplier scorecard

A red OEM number never stays at the OEM. It arrives at the tier one as commodity-level PP100 attribution, then in your monthly quality review as a directive: move the number. Now you have two moves. Fix the failure mode, or negotiate the label. One costs a €60,000 tooling modification and eight weeks of validation; the other costs an analyst a day and a fluent argument about survey methodology. I have watched the second option win that comparison more times than I can count. The arithmetic is right. The ledger is wrong.

I once sat in an OEM quality review where the defect code was negotiated harder than the corrective action. The failure was real. Our 8D was in its third iteration. We still spent two hours on whether the claim sat under a code that carried weight in their PP100 chart or one that did not. Recoding moved their internal number by a fraction of a point. The 8D closed the failure mode for good. Only one of those facts reached the vice-president's slide.

There is a better path, and I have walked it under exactly this pressure. At WITTE Automotive we ran QRQC within the shift, A3 for systemic causes and a Q-Wall that made every defect visible to everyone. Failure costs fell substantially – not because we out-argued anyone, but because there were fewer defects left to classify. Later, at SNOP, building the quality function of a 900-strong greenfield plant, we cut defect costs by 70%, held customer satisfaction at 98% and closed a quarter with zero critical customer escalations. Nobody negotiates labels in that condition. When the defect is gone, the label is a formality.

Treat initial quality as a leading indicator, not a trophy

The metric is not worthless; it is misused. PP100 tells you what customers notice in their first quarter of ownership. That is useful knowledge. What you do with it is where honesty is won or lost.

First, correlate PP100 with warranty claims by part number and failure mode. The modes that stay silent at day ninety and light up at month fourteen are your real risk register. This used to be a summer project for a black belt; today it is a weekend of data engineering. The excuse has expired.

Second, audit classification drift the way you would audit a process. Plot the monthly share of "no fault found" and "customer education" codes by analyst and by line. A drifting process has assignable causes. Usually the cause is a target someone handed down.

Third, keep latent failure modes on the same ledger. Corrosion, durability and electronics ageing belong in the same PFMEA and the same management review as the rattle reported in week three. Separate decks create separate priorities. The latent deck always loses the meeting.

A word from the boardroom side, having led quality initiatives across a 2,000-strong, multi-site workforce: never hand a survey ranking to an organisation as a target. Organisations optimise whatever you paint on the wall. Paint a rank and people will negotiate codes. Paint failure modes, owners and dates, and they will eliminate defects. If an ambition like Ford's must cascade, cascade the driver list – never the trophy.

The industry scores three months of a relationship the customer keeps for a decade. Somewhere in Dearborn a target has been set, and in supplier plants on three continents quality engineers are already weighing an analyst day against a tooling change. The customer keeps the car – through five winters, two software eras and a hundred thousand kilometres. Keep both books. The survey takes the applause in spring. The warranty bill takes the last audit, and it reads every line.