An 8D report landed on my desk detailing a field failure that had already cost a supplier significant rework charges. The root cause was a fixture alignment drift that had developed over six months on a critical machining centre. The corrective action was straightforward. The systemic question was how the drift had escaped detection when a layered Kamishibai audit specifically covered that machining centre every single day.
I pulled the historical audit data for the station. Six months of daily cards, all green. The board had not detected the drift because the board was not measuring the process. It was confirming a commercial relationship. The supplier's quality team treated the cards as customer-facing displays, and the customer's SQE treated the board as proof that standards were being met. Both parties had a strong incentive to keep the cards green, and the measurement system quietly died at the point of collection.
Across two decades implementing IATF 16949 and AS9100 systems in automotive and aerospace plants, I have seen this exact decay pattern repeat whenever an audit metric crosses a commercial boundary. The cost of that decay is rarely visible on the supplier's quality scorecard. It surfaces months later in expedited freight charges, warranty claims, and the engineering hours consumed writing corrective action reports for defects the audit system was designed to prevent.
The economics of audit theatre at the supplier boundary
A healthy Kamishibai board produces a distribution of findings. It exposes micro-failures in standard work, tool availability, and housekeeping before they compound into deliverable defects. When I reviewed this supplier's board alongside their rising PPM trend, the contrast was stark. The process was deteriorating, but the audit data was pristine. The board was generating a pleasant commercial interaction, not process intelligence.
The corruption begins with commercial awareness. Inside a single plant, a decayed card usually means a leader is going through the motions. At the supplier interface, the measurement failure is structural and economically rational. A red card is not just a process finding; it is a signal sent to the organisation that pays for the parts. Raising a red card risks triggering a chargeback, a delayed payment, or a downgrade in supplier rating that affects future contract awards.
The supplier's quality manager has a direct financial incentive to accompany the customer on the walk and manage the social environment around the data collection point. Operators give rehearsed answers. The leader conducting the audit asks fewer probing questions because aggressive verification feels adversarial. The walk generates a compliant data set while the fixture continues to drift. The cost of that drift is deferred, but it compounds daily.
How abstract criteria inflate quality costs
Working backwards from the green data, the first structural failure was card authorship. The customer had written the audit criteria in isolation. They wrote what they hoped was true rather than what could be operationally measured at the supplier's point of use. The card for the failed machining centre read 'Verify PPAP compliance.' This is a measurement with no operational definition.
The supplier's shift leader cannot assess PPAP compliance during a fifteen-minute walk. They cannot verify regulatory documentation or engineering sign-offs on the shop floor. They flipped the card green because the question was unmeasurable in context. Both parties accepted the result because neither wanted the commercial conversation a red card would trigger. The card produced administrative noise, and the alignment drift remained invisible to the audit until it surfaced as a field failure.
A valid measurement instrument must produce repeatable results regardless of who conducts the check. If the written standard cannot be physically verified by the floor leader in the time allotted, the measurement system is broken by design. The abstract reference guarantees a false positive. That false positive masks the precise failure modes the customer cares about, converting a cheap in-process correction into an expensive downstream defect.
Abstract Criteria vs. Operational Definitions at the Supplier Boundary
Cards that produce noise
- Verify PPAP compliance is maintained
- Check 5S compliance in the assembly area
- Ensure tools are properly stored
- Verify documentation is complete
Cards that produce data
- Calibration stickers on all three torque wrenches dated within 12 months
- All three shadow board positions hold the correct calibrated tool
- Last three shift logs contain all six required fields with no blanks
- Operators wear safety glasses and E-stop guard is fully down
The financial drag of delayed escalation
The second structural failure was one of escalation authority. The customer had assigned the daily Kamishibai walk to a junior supplier engineer. This engineer could record data, but lacked the authority to allocate resources or stop the line. The system detected micro-nonconformances but possessed zero corrective power. Findings were logged, routed through supervision, and arrived at a decision-maker long after the production context had vanished.
This delay is structurally amplified because the finding must cross an organisational boundary before it triggers action. The supplier's quality team had to assess whether a finding warranted formal customer notification under IATF 16949 escalation requirements. By the time both organisations had processed the paperwork, the operator who generated the relevant data had moved to a different part number. The cost of the delay is absorbed in the cost of quality as containment, rework, and sorting.
A Kamishibai board that tracks green and red cards without tracking closure velocity is measuring detection without measuring response. The walk must be structured so the supplier leader conducting it has immediate authority within the audit scope. A shift supervisor must be able to halt the process to address a finding. If escalation is required, the path must reach the supplier's quality manager within the same day. A red card that dies inside the supplier's quality system teaches both organisations that the measurement system is decorative.
A red card that dies inside the supplier's quality system teaches both organisations that the measurement system is decorative.
Surveillance budgets vs. diagnostic investment
The deepest structural failure was conceptual. The Kamishibai walk had decayed into a surveillance metric. The customer arrived looking for defects to count; the supplier braced for blame. The green card meant the customer went away satisfied. This framing guarantees the supplier will manage the audit to minimise findings. The system designed to surface failures becomes a system that actively conceals them under a layer of compliant data.
Toyota's original logic for Kamishibai applies directly to the supplier interface: the leader is not checking on the operator. The leader is checking on the system. If a supplier operator is bypassing a standard work element, the diagnostic question is what in the process makes the standard difficult to follow. The control plan may specify tools that are unavailable. The PFMEA may have missed a failure mode the process regularly exposes.
This shift requires the customer to explicitly decouple Kamishibai findings from commercial scorecard metrics. If the SQE treats every red card as a contractual dispute rather than a process measurement, the supplier will stop reporting problems. The customer must respond to red cards with engineering support, not punitive demands. Verification that leads to punishment produces concealment. Verification that leads to corrective support produces diagnostic data that prevents future cost accumulation.
Rebuilding the escalation chain as a cost-control mechanism
Fixing the system means rebuilding the escalation chain as a measured, timestamped process. A red card must trigger a sequence of events with a defined cadence and a named owner at every stage. When a card turns red, the data must flow differently than it does for routine production. The cadence must be fast enough that the finding is still technically relevant when it reaches the person who can authorise a fix.
The critical trust failure happens when red cards sit unanswered. When the supplier sees no customer response to a reported issue, they stop reporting. When the customer sees no supplier action on a flagged nonconformance, they stop trusting the board and revert to incoming inspection. Both behaviours destroy the measurement system and add inspection costs. Recovery means closing every open red card and visibly communicating the resolution back to the floor teams.
The escalation pathway must be designed to minimise the financial exposure window. The longer a nonconformance sits between detection and correction, the more expensive it becomes. A finding caught at the point of work costs minutes of production time. The same finding caught at incoming inspection costs hours of sorting. The same finding caught in the field costs weeks of warranty work and potential reputational damage.
Cross-Boundary Red-Card Escalation
- 01Immediate detectionSupplier leader investigates the nonconformance at the point of work, not at a desk
- 02Shift-level loggingIf unresolvable on the spot, assign owner and target resolution within the shift
- 0324-hour escalationSystemic issues move to supplier quality manager and customer SQE for resource allocation
- 04Adjacent-area verificationSupplier confirms the fix and audits neighbouring processes for the same failure mode
Measuring the board as a financial asset
Do not attempt to recover sixty cards across a supplier base simultaneously. Start with eight to ten cards on the supplier's most critical process, specifically the one driving the majority of nonconformance costs or PPM deviations. Define the verification criteria jointly with the supplier's process engineers. Schedule the walks and protect that time with the same priority given to a Tier 1 customer audit.
Review the rebuilt system monthly with the supplier's quality manager and treat the board as a dataset that protects margin. If a card has never turned red in six months, the verification criterion is too easy or the walk is superficial. The measurement is not reaching the process. If a card turns red every week for the same failure mode, you have a systemic process failure that requires permanent corrective action under the IATF 16949 or AS9100 nonconformance process.
You must track three specific metrics to distinguish a functioning audit system from a diplomatic exercise. Red-card frequency tells you whether the audit is actually detecting nonconformance. Closure velocity tells you whether the escalation chain is functioning. Recurrence rate tells you whether the corrective actions are effective. A board with zero frequency is a system that has failed in a way the green cards are designed to hide.
Three Metrics That Reveal Board Health