The rating is litigated. The defect isn't.

Centene put lawyers on its Medicare Advantage star ratings this month, and the reflex is understandable. In that world a star is not a score. It is a revenue line measured in billions of dollars of bonus payments. Most plants cannot afford litigation, so we run the budget version: chargeback disputes, scorecard corrections, 8Ds rewritten until the auditor stops asking questions. Ask any quality director which meeting convenes faster after a score drop – the dispute call with the customer, or containment at the machine. I have never had to wait long for the answer.

When the score carries money, the money defends the score

Stars decide bonus payments, so the ratings get a legal department. Supplier scorecards decide sourcing and approved-vendor status, so they get a commercial escalation path. VDA 6.3 and AS9100 scores decide whether the contract survives the year, so the plant builds a preparation programme that starts six months before the auditor's flight. The architecture is identical at every altitude: once the number carries money, resources flow to defending the number, not to removing the failure that feeds it. The failure modes that produced the rating wait in line. The defence of the rating does not.

FOREAST work has put me on both sides of that table. One supplier burned three engineer-months – call it €25,000 of loaded time – assembling evidence against a scorecard worth maybe €31,000 in chargebacks. The binders were immaculate. The defect that generated the entries shipped again the following Monday. Same part, same station, new shift. Nobody photographs a defect with the care that goes into photographing the paperwork.

The dispute department is fully staffed

Metric custody quietly replaced process custody as the quality department's core product, and nobody signed off on the change. We manufacture classification arguments: cosmetic or functional, whose standard applies, whether the customer's sort was justified. I have watched that single debate run four weeks. The nonconformance kept running too – four more weeks of it, in production.

Building the quality function at SNOP's greenfield plant – 900+ people, from the ground up – did not spare me scorecard wars. Every customer chargeback arrived with a rebuttal packet half-drafted out of habit. I changed one rule: no rebuttal leaves this building without a live containment attached. The customer receives the same 8D we run internally – escape point named, detection failure named, PFMEA revised, verification data promised by a date we then hit.

Some chargebacks we still disputed, on the merits, with evidence. Most simply stopped, because QRQC and A3 work at the machine cut defect cost by 70%, and a scorecard that has lost its raw material corrects itself. Within a quarter we recorded zero critical customer escalations, and I picked up a reputation as someone who de-escalates customers. I never won a single one of them by winning an argument.

Escapes write the scorecard. Appeals only write footnotes.

What an hour of appeal really costs

Nobody runs the arithmetic. A good quality engineer costs €3,000–4,000 a week loaded, and a serious chargeback defence eats two of them before anyone decides anything. Against an €18,000 claim, a "successful" dispute can burn €7,000 of engineering time and still hand the account to the competitor who contained the failure instead. The deeper cost never reaches an invoice: every hour in appeal is an hour not spent in QRQC at the machine, where the rating is actually manufactured. That is the uncomfortable framing. A rating is manufactured, defect by defect, on the line. The dispute process never touches it.

An 8D written for the auditor explains why the chargeback was disproportionate. An 8D written for the line names where the defect escaped, which control failed to catch it, what changed in the control plan, and when the verification data will exist. Same document number. Different species. The second one compounds: transparent containment teaches the customer that your first move is the machine, while a contested chargeback teaches them that your first move is defence. Trust is a stock. Every appeal is a withdrawal.

The mobilisation test

Here is an audit that costs nothing. After the next score drop, timestamp two things: when the dispute machinery convened – call scheduled, evidence owner assigned, escalation path opened – and when containment physically reached the machine that made the escape. Run it three times. If the paperwork reliably moves faster than the people, you have not built a manufacturer with a quality department. You have built a ratings-defence company that occasionally ships product. No org chart will admit this. The timestamps will.

The Centene suit is the fully funded end of a spectrum that begins with a quality engineer quietly reclassifying a nonconformance so the monthly number looks better. Same instinct, bigger budget. Across two decades in automotive and aerospace I have learned that the dispute process is the most reliably staffed process in quality – and the containment team should be. Ratings follow escapes. They have never followed appeals.