TBP A3 — Regulation Non-Conformance v3.0 Strengthened · Kalyan Critique Rubric Applied

Plant: Assembly Services Sdn Bhd (ASSB), Bukit Raja Area: Final Inspection / Quality Assurance Owner: [HOD Quality — your name] Theme ref: [A3 ref / audit ID] Method: TBP 8-Step + 5-Whys (Q5) + POO + 3D Background + Kalyan R1–R5 Status: [Draft / In-review / Approved] Date: [DD-MMM-YYYY]

0Background — Clarify the Focus (3D Split, v3.0)

New in v3.0: Step 0 is no longer a single paragraph. It is split into three lenses (Customer / External / Internal) so the team understands why this problem matters now from all sides before touching a tool. Kalyan's #1 critique of weak A3s: "team skipped Step 0 and jumped to Step 1."

⛔ Before (v2.0 — single paragraph)

"We have had several regulatory NC findings this year. Need to fix them."

Source: Typical shop-floor A3 — background reduced to one vague line.

✅ After (v3.0 — 3D split)

1. Customer (External): [e.g., JPJ type-approval audit Q2-2026, ASEAN export customer — 2 letters received, model launch at risk]

2. External (Regulation / Standard): [e.g., UN R48 / MS 1234 — headlamp leveling device marking missing on export Hilux batch; auditor cited 14 units across 3 lots]

3. Internal (Plant operation): [e.g., Final Inspection line, station 7 (PI walk); QMS doc-control gate did not cross-reference regulator update to WI revision]

Why now: [e.g., Q2 internal audit finding #ASSB-2026-Q2-014 + customer escalation letter dated 14-Apr-2026 → shipment hold risk on Hilux 4×4 export → ¥X million exposure]

⛔ Before (v2.0)

In-scope: "Regulation NCs" (vague).

✅ After (v3.0 — explicit boundary)

IN SCOPE: [e.g., All Hilux & Fortuner export models, headlamp & signal-lighting markings, all shifts, last 6 months NCs]

OUT OF SCOPE: [e.g., Other themes, non-regulatory NCs, supplier-side design changes outside ASSB control]

1Clarify the Problem — Gap = Ideal − Current

⛔ Before (v2.0)

Current: "We have NCs." / Ideal: "No NCs." / Gap: "Many."

5W1H often filled with "—" or one-liner.

✅ After (v3.0 — quantified + 5W1H)

Current: [e.g., 14 NC findings / month over last 6 months; 8 different stations; 3 different clauses cited]

Ideal: [e.g., 0 regulatory NC findings per audit cycle (parent-plant benchmark: 1.2/month)]

Gap = 14 − 0 = 14 NC/month

5W1H:

  • WHAT: [e.g., Headlamp leveling device marking missing on Hilux export — UN R48 §6.2.6.1.2]
  • WHERE: [e.g., Final Line, Station 7 — Pre-Delivery Inspection walk]
  • WHEN: [e.g., Jan–Jun 2026, all shifts, peaked in Mar after WI revision v3.2]
  • WHO: [e.g., Affected: ASEAN export customer + 14 unit-holders in field]
  • WHY: [e.g., Type-approval non-compliance → shipment hold + recall risk if fielded]
  • HOW MUCH: [e.g., 14 NC/month × 6 mo = 84 units, RM 250k rework cost, 1 audit risk score]

2Break Down the Problem — Decompose → Pareto → POO

⛔ Before (v2.0)

POO box often left empty or "Final Inspection" written without evidence.

✅ After (v3.0 — Pareto + explicit POO with gemba evidence)

Pareto (top 3 stations = 82% of NCs):

RankStationNC count (6 mo)% share
1Station 7 — Final PI3440%
2Station 5 — Sub-assy lighting2226%
3Station 3 — Trim & final finish1416%
4–5Other stations1418%

🎯 POO (Point Of Occurrence): [e.g., Station 5 — Sub-assy lighting, gemba walk 22-May-2026, led by HOD Quality + Assembly Section Head]

⚠ TBP hygiene check: POO ≠ station where NC was caught; POO = station where NC first appears. gemba evidence dated, names on it.

3Set a Target — How Much + By When + Why Challenging

⛔ Before (v2.0)

"Reduce NCs." — no number, no date, no justification of why it's not fantasy.

✅ After (v3.0 — dual-metric + challenge rationale)

BIG-TARGET: [e.g., Reduce regulatory NC from 14/month to ≤ 2/month by 30-Sep-2026 (3 months from baseline close)]

Sub-targets (decomposed by lever):

  • Station 5 (POO): [e.g., 0 NC/month from lighting sub-assy by 31-Jul-2026]
  • WI cross-ref gate: [e.g., 100% WI revisions trigger reg-diff check by 15-Aug-2026]
  • Inspector certification: [e.g., 100% Final-Line inspectors re-certified to UN R48 §6 by 31-Aug-2026]
  • Audit findings: [e.g., close all 8 open audit findings by 15-Sep-2026]

Why challenging (not BAU, not fantasy): [e.g., 18-month absence of POO analysis; lighting WI has 4-station defect flow; key gauge has no MSA; supplier on 8-week lead time for new sub-assy; requires QMS gate redesign — not a low-hanging-fruit target]

4Analyze Root Cause — 5-Whys reaches Q5 (process/system) + Rule Out Alternatives

⛔ Before (v2.0 — Kalyan's #1 critique)

5-Whys stops at Q2–Q3 and blames operator:

Why 1: NC slipped through → Why 2: inspector missed it → Why 3: operator careless. Done.

No alternatives ruled out. TBP-hygiene fail.

✅ After (v3.0 — Q5 + process/system + 5 alternatives ruled out)

Why 1: Headlamp leveling device marking missing on Hilux export units
Why 2: Inspector at Station 7 PI walk did not flag the missing marking
Why 3: PI checklist does not include the UN R48 §6.2.6.1.2 sub-clause check
Why 4: WI revision v3.2 (Mar 2026) was issued without cross-reference to the regulator update of Dec 2025
Why 5: QMS document-control gate (QP-007) does not force a reg-diff cross-reference when WIs are revised
ROOT CAUSE: QMS gate does not link WI changes to external regulation diffs → process/system cause, not operator cause.

Alternatives RULED OUT (TBP requires this):

  • Operator carelessness — ruled out: trained, signed off, multiple shifts affected, 3 inspectors on different tenure
  • Low operator motivation — ruled out: no recent change in attendance/rotation; senior inspectors involved
  • Random defect — ruled out: clustered at Station 5, repeatable on re-inspection, 3 lots affected
  • Bad supplier batch — ruled out: sub-assy from 2 different suppliers + 1 in-house, all affected
  • Equipment failure — ruled out: PI gauge MSA in spec (Apr-2026 cal), no drift detected

5Develop Countermeasures — ≥5 candidates, selection matrix, 5W2H

#CountermeasureEffectCostSpeedRecurTotalDecision
CM1Add UN R48 §6.2.6.1.2 sub-clause check to PI checklist (Station 7)HL2 wkL8✅ ADOPT
CM2Update QP-007 QMS gate to require reg-diff cross-reference on every WI revisionHM6 wkL7✅ ADOPT
CM3Re-certify all Final-Line inspectors to UN R48 §6 (2-day training + assessment)HM4 wkL7✅ ADOPT
CM4Source new lighting sub-assy with built-in marking plate (supplier design change)HH12 wkL5🟡 PARK
CM5Add 100% manual marking station post-PI (rework loop)MM2 wkH3❌ REJECT (recurrence H — does not fix root cause)

5W2H for adopted CMs (CM1, CM2, CM3):

CMWHATWHO (owner)WHEN (start→end)WHEREWHY (link to root cause)HOWHOW MUCH
CM1Update PI checklist v4.0[QA Section Head][15-Jul → 31-Jul-2026]Station 7Catches missing marking at POO downstreamChecklist rev + sign-off + 1st-piece check[RM 0 — internal]
CM2Revise QP-007 QMS gate[HOD Quality + Doc Contrl Mgr][01-Jul → 15-Aug-2026]Doc ContrlAddresses ROOT (reg-diff link)Gate redesign + pilot on 3 WIs + roll-out[RM 5k consulting]
CM3Re-certify 18 inspectors[Training Section Head][01-Aug → 31-Aug-2026]Training roomCapability uplift = human-side defense2-day class + written + practical test[RM 18k — internal cost]

6+7Implement (Do) + Monitor (Check) — 3-perspective

Step 6 — Implementation Schedule

CMOwnerStartEndQ-by (review checkpoint)Status% complete
CM1 — PI checklist v4.0[QA Section Head]15-Jul31-Jul07-Aug[On-track / At-risk / Done][%]
CM2 — QP-007 gate[HOD Quality + Doc Contrl]01-Jul15-Aug22-Aug[...][%]
CM3 — Inspector cert[Training Section Head]01-Aug31-Aug07-Sep[...][%]

Step 7 — Monitor Results AND Processes (3-perspective)

⛔ Before (v2.0)

"Monitor and review monthly." — no metric, no owner, no 3-perspective split.

✅ After (v3.0 — 3-perspective split)

👤 CUSTOMER perspective (Voice of regulator / external customer): [e.g., 0 new findings on Jul/Aug/Sep QMS audit; ASEAN customer letter of closure received]

🏢 TOYOTA perspective (QMS, parent plant, brand): [e.g., Zero recurrence for 90 days post-CM3; QP-007 gate live across all WI revisions; parent plant benchmark matched (≤1.2 NC/mo)]

🧑 YOUR OWN perspective (Team learning, capability): [e.g., 5-Whys depth proven at Q5 by team independently on next A3; gemba frequency up from 1×/mo to 2×/mo; 2 inspectors passed certifying-exam with distinction]

⚠ Contingency Plan (if CMs miss target): [e.g., If NC still >4/month by 15-Sep → escalate to Plant Director, trigger CM4 (supplier design change) early, second-line CM = in-line auto-vision check at Station 5]

8Standardize + Yokoten (success AND failure modes) + Reflection

⛔ Before (v2.0)

Yokoten = "share with sister plant" (slogan, no date, no mechanism).

✅ After (v3.0 — include failure modes + mechanism + date)

New / Updated Standards:

  • [WI-PI-007 v4.0 issued 31-Jul-2026 — UN R48 §6.2.6.1.2 sub-clause check added]
  • [QP-007 v2.0 issued 15-Aug-2026 — reg-diff cross-reference gate]
  • [TRN-2026-FIN-03 — UN R48 §6 cert added to Final-Line role matrix]
  • [DCR-2026-087 — QMS gate change record]

Yokoten (horizontal deployment — share BOTH success and failure modes):

  • Deploy to: [e.g., All 3 ASSB lines (A, B, C) + sister plant ASSB-Simpang Ampat by 31-Oct-2026]
  • Mechanism: [e.g., QCC bulletin #ASSB-Q3-2026-009 + cross-functional meeting 15-Oct + gemba visit to sister plant 22-Oct]
  • Failure modes to share (per Kalyan R5 — anti-patterns to avoid):
    • ❌ Don't stop 5-Whys at Q2 and blame operator
    • ❌ Don't use "Final Inspection" as POO without gemba evidence
    • ❌ Don't pick vague target like "reduce NCs" — needs number + date
    • ❌ Don't issue WI revisions without reg-diff cross-reference (this is what caused the original NC)

Next Kaizen — climb the PDCA ladder:

  • Level 1 (this A3): Closed regulatory NC at POO Station 5
  • Level 2: [e.g., Standardised across all 3 lines, all shifts, by 31-Oct-2026]
  • Level 3: [e.g., Built into QMS gate — auto cross-ref of reg updates, by 31-Dec-2026]
  • Level 4: [e.g., KPI: 0 NC for 12 months → target next theme: emissions compliance Q4-2026]

🪞 Reflection (v3.0 Mandatory)

New in v3.0: Reflection is no longer a single line. It is a structured self-evaluation in 4 parts — what worked, what surprised, what I would do differently, patterns to fix. This is the only tab a Plant Director reads twice.

🤖 BotBot's self-evaluation on this strengthened A3

1. What worked (keep doing):

2. What surprised me:

3. What I would do differently next time:

4. Patterns to fix (3× = bias to MEMORY):

Kalyan Critique Rubric R1–R5 — Self-Check

From Case Study #2 (Kalyan TMA-SG, 2026): An A3 is "ok" if it fills the template. It is defensible if it passes R1–R5. Apply before submitting to Plant Director.

R1
Type of problem — Setting / Reaching / Trouble? Each has different CM idiom. State it explicitly.
✓ Trouble (NC found at audit, must fix & prevent recurrence)
R2
Countermeasures labeled as Defensive / BR (break recurrence) / BR-momentum. The mix tells the story.
✓ CM1=Defensive, CM2=BR, CM3=BR-momentum — balanced
R3
Owner name + Q-by date for every CM. No "team" / "TBD" / "ongoing".
✓ Every row has name + date placeholders ready
R4
Pilot study + 26APF input connection — show how the deck is a plan, not a report.
✓ QP-007 pilot on 3 WIs → 26APF roll-out path defined
R5
Reflection in TBP voice — what worked / what surprised / what next — not consultant-speak.
✓ 4-part structure above; pattern tracking explicit

Legend — What's New in v3.0

Yellow placeholder = fill with real ASSB data Green = new v3.0 field (not in v2.0 template)