Kobetsu KaizenJapanese 'individual / focused improvement' — JIPM TPM Pillar 2

Kobetsu kaizen — literally 'individual / focused improvement' — is the second pillar of the eight-pillar JIPM Total Productive Maintenance model. While autonomous maintenance (Pillar 1) builds the daily operator-led discipline of clean-inspect-lubricate-tighten and standard kaizen (the company-wide everyday-improvement practice) handles the long tail of small frictions, kobetsu kaizen is the chartered, cross-functional, time-boxed attack on a single chronic loss that resists routine fixes. The unit of work is a project, not a suggestion.

Two characteristics distinguish kobetsu kaizen from everyday kaizen. First, the team is cross-functional by design — operator + maintenance technician + process engineer + quality engineer + planner — because chronic losses sit at the seams between functions and no single function owns the data needed to solve them. Second, the analysis is rigorous: PM analysis (Phenomena-Mechanism analysis), why-why to 5 levels minimum, fishbone with 4M+E branches, and validated root causes — not hunches. The output is a documented countermeasure, a trial, a verified result, a standardised work / equipment change, and a horizontal-deployment plan (yokoten) to every similar line.

Nakajima's original 6 big losses became 8 equipment losses in the 1990s and then the full 16-loss model in Suzuki's process-industry treatment. Without the loss taxonomy a kobetsu-kaizen team picks themes from gut feel; with it, the team targets the 2–4 losses that account for 60–80% of the gap to world-class OEE.

The team starts by running a 4-week loss survey: every stop > 1 minute, every defect, every changeover, every speed drop, every motion-wasted minute is logged to one of the 16 losses. The resulting Pareto almost always shows 2–4 losses dominating. That is the kobetsu kaizen backlog for the next 12 months.

Every JIPM-trained plant runs kobetsu kaizen on the same 10-step backbone. It is PDCA at project scale with explicit checkpoints for theme selection, root-cause validation, horizontal deployment and sustain — the four steps weak kobetsu-kaizen practices typically skip.

1. Theme selection. Pick one loss from the 16-loss Pareto. Score candidates on: cost impact (£/year), feasibility, strategic alignment (hoshin link), team availability. Chartered, named, signed by the area manager. Charters that drift to 'whatever the team finds' produce no result.
2. Understand the current condition. Go to the gemba. Measure with a stopwatch, a data logger, a thermal camera, an oscilloscope — whatever the loss demands. Quantify in the units the loss is paid in (minutes lost, units scrapped, kWh wasted). 5 sigma of fact before any analysis.
3. Set the target. Specific, measurable, time-bound. 'Reduce minor stops on station 4 from 14.2 events/hr to ≤ 2 events/hr by end of month 4.' Targets that are not numbers are wishes.
4. Analyse the loss. PM analysis is the canonical method — define the phenomenon precisely, list every physically possible mechanism, eliminate by evidence. 5 Whys / why-why to 5 levels minimum. Fishbone (4M+E). Engineering analysis (FEA, tolerance stack, fluid dynamics) where the mechanism is non-obvious. Validate root causes — do not assume.
5. Plan countermeasures. Each countermeasure traces to a validated root cause. Score by effectiveness × feasibility × cost. Choose 1–3 to trial.
6. Trial / implement. Controlled trial first — 1 station, 1 shift, 1 week. Measure against baseline. Adjust. Roll out fully only when the trial proves the effect.
7. Confirm the effect. Compare actual against target. Statistical-significance test where applicable (paired t-test on shift-level data, control chart). Did we hit the number? If no, return to step 4 — the root cause was wrong.
8. Standardise the gain. Revise the standard work, the equipment drawing, the PM schedule, the operator training. Lock the new state into the document-control system + training-record system + change-control system. Without this step gains evaporate within 90 days.
9. Horizontal deployment (yokoten). Identify every other line, station, or piece of equipment with the same mechanism. Deploy the countermeasure to each. This is where the OEE multiplier lives — one validated countermeasure can lift OEE on 8 lines at the cost of designing it once.
10. Sustain + next theme. 30/60/90-day sustain audits prove the gain held. Close the project with an A3. Select the next theme from the loss Pareto. The cycle is continuous; mature plants run 12–20 kobetsu-kaizen projects per area per year.

All three belong in a mature plant. Everyday kaizen without kobetsu kaizen plateaus once the easy frictions are gone — operators run out of fixes they can implement alone. Kobetsu kaizen without everyday kaizen attempts to fix every small problem with a chartered cross-functional project, which is too slow and too expensive. Kaikaku without either is a one-off step-change that decays back to the old OEE within a year because nobody is improving the new state.

PM analysis (Phenomena-Mechanism analysis) is JIPM's structured method for problems where 5 Whys runs out at why-3 because the team genuinely does not know the physical mechanism. Where 5 Whys assumes the answer is one chain of reasoning, PM analysis enumerates every physically possible mechanism and eliminates them by evidence.

1. State the phenomenon precisely. Not 'the machine jams' but 'between gram weights 0.4–0.6g, the reject chute jams approximately every 6 minutes; clearance is achieved by manual prod within 8s; no foreign material is present.'
2. Decompose physically. List every physically possible mechanism — friction, geometry, vibration, electrostatic, humidity, material property, control-logic timing, sensor drift, operator action.
3. Identify required conditions. For each mechanism, what physical conditions must be present for that mechanism to occur? E.g. 'electrostatic attraction requires non-conductive product + dry air < 35% RH + chute material with high tribocharge ranking.'
4. Survey current conditions. Measure every required condition. Is RH < 35%? Is the chute high-tribocharge material? Is the product non-conductive?
5. Identify the mechanisms supported by evidence. Eliminate those whose required conditions are not present. Typically 1–3 mechanisms remain.
6. Plan countermeasures against the remaining mechanisms. Test the countermeasures. The remaining mechanism after testing is the validated root cause.

PM analysis is most useful for minor-stops + speed-loss + defect-mode problems where intuition has failed for years. Plants that adopt it find that chronic losses long-thought-unsolvable (the kind that have lived in the loss report for a decade) yield within one or two kobetsu-kaizen cycles.

In regulated manufacturing kobetsu kaizen is the loss-elimination discipline that produces the auditable continual-improvement evidence regulators expect — but it must run inside the regulated wrapper, not parallel to it.

• ICH Q10 §3.2.4 — continual improvement of process performance and product quality. Kobetsu-kaizen project portfolios and 30/60/90-day sustain results are the strongest evidence under §3.2.4. Plants without a visible kobetsu-kaizen pipeline get 'continual improvement program inadequate' findings.
• 21 CFR 820.100 / ISO 13485 §8.5.2 — CAPA. A kobetsu-kaizen project triggered by a quality loss (defects/rework/yield) IS a CAPA when the loss meets the CAPA threshold. Run them as one workflow — the JIPM 10-step doubles as the CAPA investigation + effectiveness verification.
• 21 CFR 211.100 / 820.30(i) — change control. Any countermeasure that changes a validated state (equipment parameter, in-process control, BOM, work instruction) routes through change control before deployment. The horizontal-deployment step is a change-control batch action, not a free-for-all.
• ICH Q9(R1) — quality risk management. Each kobetsu-kaizen countermeasure carries a risk assessment: does eliminating this loss introduce a new failure mode? Documented in the project A3, signed by quality.
• 21 CFR 211.180(e) — APR / PQR. Kobetsu-kaizen results are summarised in the annual product review as the 'process performance improvements' section.
• 21 CFR 820.20 / ICH Q10 §2.7 / ISO 13485 §5.6 — management responsibility. Kobetsu-kaizen project portfolio + completion rate + financial impact is a standing management-review input. Visible top-management ownership is what auditors look for.
• EU GMP Annex 15 / ICH Q12 — qualification + lifecycle change management. Changes to qualified equipment may trigger re-qualification (often partial, justified by risk). Build the re-qual trigger into the project plan, not as a surprise at the end.

• Active project count per area — healthy: 2–4 concurrent. < 2 means the discipline is dormant; > 6 means resources are spread too thin to deliver any.
• Completion rate within charter — % of projects that hit target within charter duration. World-class: 70%+. Below 50% means theme-selection or root-cause analysis is weak.
• Loss-Pareto coverage — % of the loss Pareto value attacked by an active or completed project. Mature plants: 70%+ within 18 months.
• OEE impact attributable to kobetsu kaizen — measured at area level pre vs post. World-class plants attribute +15–25 OEE points over 3 years to kobetsu kaizen.
• Horizontal-deployment ratio — countermeasures deployed beyond originating line ÷ countermeasures developed. World-class: 2–4× (every solution reused on 2–4 other lines). < 1.2× means the yokoten step is being skipped.
• Sustain rate at 90 days / 180 days / 365 days. World-class: 90% / 85% / 75%. Below means the standardise-the-gain step is not landing in document control + training + PM schedule.
• Financial benefit realised vs forecast — for capital-justified projects. World-class: actuals within 80–120% of forecast within 12 months of completion.
• Quality-loss reduction — defects-per-million / scrap rate / first-pass yield at area level. The most consequential output for regulated plants.

1. No loss Pareto. Teams pick themes from gut feel; the projects do not move OEE because they were not on the critical path. Always start with the 4-week loss survey.
2. Composite-KPI theme. 'Improve OEE' or 'reduce downtime' is not a theme. The discipline is loss-specific — one named loss with one named mechanism.
3. Skipping PM analysis when 5 Whys runs out. Teams default to gut-feel countermeasures and trial them blindly. The countermeasure works for a week and the loss returns.
4. Skipping horizontal deployment. The team celebrates the project on the originating line and never deploys to the other 5 lines with the same mechanism. The yokoten step is where the financial leverage lives.
5. Skipping the standardise step. The countermeasure is implemented but not captured in standard work, equipment drawings, PM schedule, training records or change control. New operators / new shifts do not know about it; the gain erodes inside 90 days.
6. Treating kobetsu kaizen as 'kaizen events.' Kaizen events are a related-but-different practice (3–5 day rapid-improvement events). Kobetsu kaizen is a 60–120-day funded project with rigorous PM analysis. Mature plants run both — but not as one workflow.
7. Bypassing the QMS in regulated plants. 'It's just an improvement' — until the inspector asks for the change-control record and the validation impact assessment. Run the JIPM 10-step inside the QMS, not parallel to it.

V5 ships kobetsu kaizen as a chartered-project workspace that wires JIPM's 10-step discipline directly into the loss-data pipeline, the regulated wrapper (CAPA + change control + validation) and horizontal-deployment automation. The platform makes the discipline cheap enough to sustain at scale — and prevents the four steps weak practices skip (theme selection, root-cause validation, yokoten, sustain) from being skipped at all.

• Live 16-loss Pareto per area. Loss data flows from kiosk + andon + OEE + quality automatically into the 16-loss taxonomy. The Pareto refreshes weekly and ranks themes by cost-per-year × confidence.
• Project charter workflow. Theme selection → scoring → charter → area-manager e-sig. Charter is the single source of truth for scope, target, team, duration, sponsor.
• JIPM 10-step canvas. Each project moves through steps 1→10 with required artefacts at each step. The project cannot close until standardise + horizontal-deploy + sustain are signed.
• PM analysis template. Structured phenomenon → mechanism → required-conditions → evidence-survey → validated-mechanism workflow with inline data attachment.
• Root-cause validation gate. Step 4 cannot close until the validated mechanism is signed by an engineer + a quality reviewer. Stops the 'gut-feel countermeasure' anti-pattern.
• Auto-route to CAPA / change control / validation. Quality-loss themes auto-create a linked CAPA; countermeasures touching validated state auto-create a linked change order with risk-assessment template.
• Horizontal-deployment engine. Once a countermeasure is validated, the platform identifies every other line / station with the same equipment + same loss signature and creates a deployment task per target with its own change-control sub-record.
• Sustain audit scheduler. 30/60/90/180/365-day sustain audits are auto-scheduled at project close; results feed the sustain-rate dashboard; failed audits auto-open a re-investigation.
• Financial-benefit tracking. Forecast at charter; actual measured monthly from KPI data; variance surfaced to project sponsor + management review.
• Management-review dashboard. Active projects / completion rate / loss-Pareto coverage / OEE delta / financial benefit / sustain rate — one screen per area, rolled up plant + enterprise.
• Regulated audit pack. One-click export per project: charter + 10-step canvas + PM analysis + countermeasures + trial data + change-control record + validation impact + training-record evidence + sustain audits — all Part 11 + Annex 11 stamped.
• Mobile-safe at iPhone width. The entire workflow renders cleanly at ≤ 390px so engineering, maintenance and operators can update from the line.