Implementation Governance and Reference Guide

Success factors, failure modes, regulatory considerations, research outlook, references, evidence summaries, maturity rubric, and glossary.

9 Critical Success Factors and Failure Modes

9.1 What the evidence identifies as decisive

Three independent literatures converge on the same conclusion, which gives it unusual weight. Netland (2016), surveying 432 practitioners across 83 factories, found management support and commitment among the highest-ranked critical success factors, with the ranking largely invariant across corporation, factory size, implementation stage and national culture. Bortolotti, Boscari and Danese (2015) found that soft lean practices, not hard ones, discriminate successful from unsuccessful implementations. Anand, Ward, Tatikonda and Schilling (2009) found that continuous improvement functions as a dynamic capability only when supported by a comprehensive organisational infrastructure covering project selection, resource allocation, structure and knowledge capture.

These are three different research designs, in three different literatures, reaching the same finding: the technical content of operational excellence is not the limiting factor. Organisations that have failed at improvement have almost never failed for want of technique.

A fourth literature converges from a different direction. The kaizen event research programme of Farris, Van Aken, Doolen and Worley (2009), covering 51 events across six organisations, and Glover, Farris, Van Aken and Doolen (2011), extending to 65 events across eight organisations, found that the factors predicting whether event outcomes sustain are organisational and social — goal clarity, team composition and autonomy, management support, and the follow-through mechanism — rather than technical. Four independent research programmes, four different designs, one conclusion.

9.2 Failure modes, with diagnostics

Failure mode	Presentation	Diagnostic question
Improvement capacity not protected	Projects stall in month three; the same people are named on every charter; improvement work is the first casualty of a bad week	How many improvement hours were committed this quarter, and how many were actually delivered? If this cannot be answered, the capacity is notional.
Measurement system unassessed	Projects chase causes that do not exist; capability figures move without process change; disagreement between shifts on the same parts	What is the attribute agreement of the inspection system that generates our quality numbers? When was it last quantified?
Hard bundle without soft bundle	Full technical apparatus in place; performance flat; improvement activity confined to a specialist function	How many hours of protected problem-solving time did front-line personnel receive last month?
Local efficiency defeating subordination	Non-constraints run full; work-in-process accumulates; the constraint still starves	Does any manager have a utilisation or absorption target on a non-constraint resource?
Technology ahead of process	Dashboards proliferate; models are built and not used; data projects overrun	Was the process in statistical control before the model was built? What decision does this system change?
Change-control queue saturation	Improvement ideas accumulate; validated changes take quarters; engineers spend the majority of their time on documentation	What is the median elapsed time from change request to implementation, and what is the current queue length?
Benefit inflation	Reported savings exceed observable financial improvement by a large factor	What proportion of claimed benefit was finance-validated at 180 days?
Composite metric management	Overall equipment effectiveness improves while output does not; cost of poor quality falls while scrap volume does not	Has any definition, boundary or exclusion in the metric changed during the reporting period?
Model drift unmonitored	A predictive system performs well at deployment and degrades silently	What is the model’s current performance against its qualification baseline, and who monitors it?
Coach scarcity masked by training volume	Large numbers trained; vocabulary widespread; behaviour unchanged	How many people can coach a structured experimental cycle without notes? Not: how many have been trained?
Late detection	Problems surface in weekly metrics rather than within the shift; the same issue is discovered repeatedly by different people	When something goes wrong on this line, how long before someone whose job it is to fix it knows? If the answer is shifts, the constraint is at L0.
PDCA in name only	Improvement work is planned once, implemented at full scale, and reported as complete	How many cycles did the last five problems take, and was the countermeasure tested on one machine before all four?
A3 as a form	A3 templates in use; each completed by one person and filed	How many drafts does a typical A3 go through before acceptance? If the answer is one, no teaching occurred.
Weak countermeasures	Corrective actions close on time; the same failure modes recur within the year	What share of the last fifty corrective actions were retraining or procedure revision? What is the 12-month recurrence rate?
Root cause asserted	Investigations complete quickly and read plausibly; recurrence is high	For the last serious problem, was the cause shown to turn the effect on and off, or was it agreed to be likely?
Skills matrix as decoration	Matrix exists and is colourful; staffing decisions never reference it	Has this matrix ever changed a shift assignment, or appeared in a risk discussion?

Table 9.1 Failure modes with field diagnostics. The diagnostic questions are designed to be answerable in a single conversation on the floor.

9.3 The specific pathology of regulated environments

Device plants exhibit a characteristic failure that unregulated plants do not: improvement effort displaced into documentation. Because the quality system requires evidence, and because evidence is easier to produce than change, organisations under pressure converge on producing evidence of improvement rather than improvement. The signature is a rising count of closed corrective actions with a flat defect rate — a pattern that is straightforward to detect and rarely looked for.

The mechanism is visible in the countermeasure profile. Retraining and procedure revision — ranks 5 and 6 of Table 4.2 — are the two cheapest actions to document and the two least likely to work, and they dominate corrective action portfolios in most regulated plants. Peerally and colleagues (2017) identify precisely this pattern in healthcare: investigations whose action plans generate weak or poorly designed risk controls. The recurrence that follows is not an analysis failure but a countermeasure-design failure, and it is measurable in any plant this week.

The countermeasure is to measure effectiveness rather than closure. A corrective action is effective when the failure mode’s rate falls and stays down, verified at a defined interval with a defined statistical criterion. Under both ISO 13485 and the QMSR this is already required in substance; the difference between plants is whether effectiveness checks are performed as a genuine test with a defined acceptance criterion, or as a documentation step. That difference is visible in the data within two quarters.

10 Regulatory and Validation Considerations

10.1 The QMSR transition and what it changes

The FDA Quality Management System Regulation took effect on 2 February 2026, amending 21 CFR Part 820 to incorporate ISO 13485:2016 by reference in place of the former Quality System Regulation, and harmonising the United States framework with that used by other regulatory authorities. FDA concurrently moved device inspections onto Compliance Program 7382.850, retiring the previous inspection programmes.

For a firm already operating a conformant ISO 13485:2016 system the substantive change is modest and largely documentary and inspectional. For a firm operating a legacy QSR-shaped system it is structural. In either case the operational excellence consequence is the same: the process-based, risk-based architecture of ISO 13485 is more hospitable to the methods in this review than the prescriptive architecture it replaces, because process capability evidence, statistical control evidence and continued process verification data map naturally onto its expectations. Firms that have built the measurement architecture of Section 8 are, in practice, better positioned at inspection than firms that have built additional procedures.

10.2 Validated processes and the change-control tax

Under ISO 13485 clause 7.5.6, processes whose output cannot be verified by subsequent monitoring or measurement must be validated. In lens manufacture this typically captures moulding, hydration, surface treatment, sealing and sterilisation — that is, most of the process chain. The consequence, developed in Section 2.2, is that improvement velocity is bounded by change-control throughput.

Three structural responses are available, in increasing order of value.

Risk-based change classification. A documented, defensible scheme that distinguishes changes requiring revalidation from those requiring verification only, applied consistently and justified against the risk file. This does not reduce the work on significant changes; it prevents insignificant changes from consuming the same effort.
Design-space validation. Validating a characterised operating region rather than a point setting, so that movement within the region is an operational decision rather than a change event (4.5). This is the highest-leverage structural response available and it is under-used in the device sector relative to pharmaceuticals.
Process-mined change control. Applying 4.14 to the change-control process itself, targeting the rework loops and handover waits that typically dominate its elapsed time. This raises the throughput ceiling for every other method in the portfolio and is usually the fastest-returning project available to a device plant.

10.3 Software, models and computer software assurance

Any software that influences product quality decisions falls within the plant’s software validation obligations. Machine-learning models introduce three complications that conventional software validation was not designed for.

Non-determinism in development. Model behaviour depends on training data, random initialisation and hyperparameters. Reproducibility requires versioning of data, code and model artefact together, not code alone.
Performance drift in use. A validated model degrades as the process changes. Validation must therefore include ongoing performance monitoring with defined thresholds and a defined retraining trigger, and retraining must itself be a controlled change.
Explainability at inspection. Where a model influences product disposition, the firm must be able to explain the basis of a specific decision. This is a strong practical argument for interpretable models and for advisory-mode deployment in which a competent human makes the recorded decision.

A workable classification for planning purposes: models that trigger engineering investigation carry low validation burden; models that inform a human disposition decision carry moderate burden and require competence controls on the decision-maker; models that make an automated disposition decision carry full validation burden and should be entered only where the economic case is strong and the performance evidence is prospective. Most of the benefit identified in Sections 4.9 to 4.11 is available in the first two categories.

10.4 Risk management as the project selection engine

ISO 14971 requires a maintained risk management file with identified hazards, estimated risks and implemented controls. That file is the most defensible project-selection instrument a device plant possesses, and it is routinely under-used for this purpose. Selecting improvement projects by ranked residual risk, rather than by cost alone, produces a portfolio that is simultaneously the highest-value operationally and the most straightforward to justify at inspection. Where a manufacturing process is a risk control, improving its capability is a risk-reduction activity and should be documented as such — which also means the improvement benefits from the risk file’s existing analytical work rather than duplicating it.

11 Research Gaps and Outlook

Five gaps are worth flagging, both because they bound what can honestly be claimed and because they indicate where a well-instrumented plant could make a genuine contribution.

Absence of controlled evidence for digitally enabled methods. Zero Defect Manufacturing, predictive quality and digital twins are supported by systematic reviews of conceptual and case literature, not by controlled comparison. There is no equivalent, for any of them, of the Swink and Jacobs matched-sample design. Until there is, adoption decisions rest on mechanism plausibility rather than on measured effect.
Economic evaluation methods. The Zero Defect Manufacturing reviews explicitly identify economic evaluation as a gap, and the predictive maintenance reviews report limited and highly variable quantitative guidance on cost-benefit. Practitioners are therefore building business cases without an accepted method, which is one reason realised benefits so often diverge from projected ones.
Regulated-environment method adaptation. Almost all method development and evaluation occurs in unregulated settings. The change-control tax, which dominates improvement economics in device and pharmaceutical manufacture, is essentially absent from the operational excellence literature. McGrane and colleagues (2022) is a rare and valuable exception, and the field would benefit from more work of exactly that kind.
Reinforcement learning verification. The comparative survey literature identifies reliability, distribution-shift robustness and verification as open problems for learned scheduling policies. In regulated manufacture, verification is not a research nicety but a deployment precondition, and no accepted approach exists.
Capability measurement. The methods with the largest plausible long-run effect — coaching, problem-solving capability, socio-technical design — are the hardest to measure, and consequently the least well evidenced and the most vulnerable in resource competition. A validated instrument for organisational problem-solving capability would change how these investments are justified.

Two directional expectations follow. The centre of gravity in operational excellence research has moved decisively toward integration frameworks — LSS4.0, DMAIC 4.0, Quality 4.0 maturity models — and away from single-method studies, which is appropriate given that firms adopt bundles rather than methods. And the human-centric turn associated with Industry 5.0, whatever one makes of the label, is recovering a socio-technical proposition that the large-sample lean literature established two decades ago and that the Industry 4.0 wave partially obscured. Engineers evaluating that turn should ground it in Bortolotti and Netland rather than in the Industry 5.0 corpus, where the evidence is thinner than the enthusiasm.

A closing methodological caution

Mabin and Balderstone (2003) reported that they could find no published accounts of failure across a large search of Theory of Constraints applications. That finding generalises across this entire literature. Published improvement results are drawn from a population in which failures do not appear, and the resulting effect estimates are upper bounds. The appropriate practitioner response is not scepticism about the methods, which are sound, but scepticism about the magnitudes — and a commitment to measuring one’s own baseline and one’s own result with more rigour than the literature applies to itself.

12 References

All entries below were checked against a publisher, repository or indexing record during preparation. Digital object identifiers are given where they were directly confirmed; where a DOI was not confirmed, the entry gives journal, volume, issue and pages only.

Peer-reviewed journal articles

Anand, G., Ward, P. T., Tatikonda, M. V. and Schilling, D. A. (2009). Dynamic capabilities through continuous improvement infrastructure. Journal of Operations Management, 27(6), 444–461.

Antony, J. and colleagues (2022). Quality 4.0 conceptualisation and theoretical understanding: a global exploratory qualitative study. The TQM Journal.

Bortolotti, T., Boscari, S. and Danese, P. (2015). Successful lean implementation: organizational culture and soft lean practices. International Journal of Production Economics, 160, 182–201. https://doi.org/10.1016/j.ijpe.2014.10.013

Bateman, N., Philp, L. and Warrender, H. (2016). Visual management and shop floor teams: development, implementation and use. International Journal of Production Research, 54(24), 7345–7358.

Card, A. J. (2017). The problem with “5 whys”. BMJ Quality & Safety, 26(8), 671–677.

Carvalho, A. M., Sampaio, P., Rebentisch, E., Carvalho, J. Á. and Saraiva, P. (2023). Operational excellence, organizational culture, and agility: bridging the gap between quality and adaptability. Total Quality Management & Business Excellence. https://doi.org/10.1080/14783363.2023.2191844

Chunhachatrachai, P. and Lin, C.-Y. (2023). CLensRimVision: a novel computer vision algorithm for detecting rim defects in contact lenses. Sensors, 23(23), 9610. https://doi.org/10.3390/s23239610

Farris, J. A., Van Aken, E. M., Doolen, T. L. and Worley, J. (2009). Critical success factors for human resource outcomes in Kaizen events: an empirical study. International Journal of Production Economics, 117(1), 42–65.

Glover, W. J., Farris, J. A., Van Aken, E. M. and Doolen, T. L. (2011). Critical success factors for the sustainability of Kaizen event human resource outcomes: an empirical study. International Journal of Production Economics, 132(2), 197–213.

Hoerl, R. W. and Snee, R. D. (2012). Statistical engineering: six decades of improved process and systems performance. Quality Engineering, 24(2). https://doi.org/10.1080/08982112.2012.654324

Hoerl, R. W. and Snee, R. D. (2014). Statistical engineering and variation reduction. Quality Engineering, 26(1). https://doi.org/10.1080/08982112.2013.846069

Hopp, W. J. and Van Oyen, M. P. (2004). Agile workforce evaluation: a framework for cross-training and coordination. IIE Transactions, 36(10), 919–940.

Kritzinger, W., Karner, M., Traar, G., Henjes, J. and Sihn, W. (2018). Digital twin in manufacturing: a categorical literature review and classification. IFAC-PapersOnLine, 51(11), 1016–1022. https://doi.org/10.1016/j.ifacol.2018.08.474

Linderman, K., Schroeder, R. G., Zaheer, S. and Choo, A. S. (2003). Six Sigma: a goal-theoretic perspective. Journal of Operations Management, 21(2), 193–203. https://doi.org/10.1016/S0272-6963(02)00087-6

Lu, Y., Zheng, H., Chand, S., Xia, W., Liu, Z., Xu, X., Wang, L., Qin, Z. and Bao, J. (2022). Outlook on human-centric manufacturing towards Industry 5.0. Journal of Manufacturing Systems, 62, 612–627. https://doi.org/10.1016/j.jmsy.2022.02.001

Mabin, V. J. and Balderstone, S. J. (2003). The performance of the theory of constraints methodology: analysis and discussion of successful TOC applications. International Journal of Operations & Production Management, 23(6), 568–595.

Maia, L. C., Lizarelli, F. L. and Gambi, L. D. N. (2024). Industry 4.0 and Six Sigma: a systematic review of the literature and research agenda proposal. Benchmarking: An International Journal, 31(3), 1009 ff. https://doi.org/10.1108/BIJ-05-2022-0289

McGrane, V., McDermott, O., Trubetskaya, A., Rosa, A. and Sony, M. (2022). The effect of medical device regulations on deploying a Lean Six Sigma project. Processes, 10(11), 2303. https://doi.org/10.3390/pr10112303

Netland, T. H. (2013). Exploring the phenomenon of company-specific production systems: one-best-way or own-best-way? International Journal of Production Research, 51(4).

Netland, T. H. (2016). Critical success factors for implementing lean production: the effect of contingencies. International Journal of Production Research, 54(8), 2433–2448.

Peerally, M. F., Carr, S., Waring, J. and Dixon-Woods, M. (2017). The problem with root cause analysis. BMJ Quality & Safety, 26(5), 417–422. https://doi.org/10.1136/bmjqs-2016-005511

Psarommatis, F., May, G., Dreyfus, P.-A. and Kiritsis, D. (2020). Zero defect manufacturing: state-of-the-art review, shortcomings and future directions in research. International Journal of Production Research, 58(1), 1–17. https://doi.org/10.1080/00207543.2019.1605228

Psarommatis, F., May, G. and Azamfirei, V. (2024). Zero defect manufacturing in 2024: a holistic literature review for bridging the gaps and forward outlook. International Journal of Production Research. https://doi.org/10.1080/00207543.2024.2388217

Roche, M. and colleagues (2025). Corporate sustainability strategy deployment: a case study on the implementation of corporate sustainability using hoshin kanri. Corporate Social Responsibility and Environmental Management. https://doi.org/10.1002/csr.2959

Sader, S., Husti, I. and Daróczi, M. (2022). A review of quality 4.0: definitions, features, technologies, applications, and challenges. Total Quality Management & Business Excellence.

Shah, R. and Ward, P. T. (2003). Lean manufacturing: context, practice bundles, and performance. Journal of Operations Management, 21(2), 129–149. https://doi.org/10.1016/S0272-6963(02)00108-0

Shah, R. and Ward, P. T. (2007). Defining and developing measures of lean production. Journal of Operations Management, 25(4), 785–805. https://doi.org/10.1016/j.jom.2007.01.019

Skalli, D., Cherrafi, A., Charkaoui, A., Chiarini, A., Shokri, A., Antony, J., Garza-Reyes, J. A. and Foster, M. (2025). Integrating Lean Six Sigma and Industry 4.0: developing a design science research-based LSS4.0 framework for operational excellence. Production Planning & Control, 36(8). https://doi.org/10.1080/09537287.2024.2341698

Steiner, S. H., MacKay, R. J. and Ramberg, J. S. (2008). An overview of the Shainin System™ for quality improvement. Quality Engineering, 20(1), 6–19. https://doi.org/10.1080/08982110701648125

Taylor, M. J., McNicholas, C., Nicolay, C., Darzi, A., Bell, D. and Reed, J. E. (2014). Systematic review of the application of the plan–do–study–act method to improve quality in healthcare. BMJ Quality & Safety, 23(4), 290–298. https://doi.org/10.1136/bmjqs-2013-001862

Tezel, A., Koskela, L. and Tzortzopoulos, P. (2016). Visual management in production management: a literature synthesis. Journal of Manufacturing Technology Management, 27(6), 766–799.

Swink, M. and Jacobs, B. W. (2012). Six Sigma adoption: operating performance impacts and contextual drivers of success. Journal of Operations Management, 30(6), 437–453. https://doi.org/10.1016/j.jom.2012.05.001

Anon. (2021). Implementation of POLCA-integrated QRM framework for optimized production performance — a case study. Sustainability, 13(6), 3452. https://doi.org/10.3390/su13063452

Systematic reviews and methodological sources

Anon. (2019). A systematic literature review of machine learning methods applied to predictive maintenance. Computers & Industrial Engineering.

Godinho Filho, M. and colleagues (2017). The application of Quick Response Manufacturing practices in Brazil, Europe and the USA: an exploratory study. International Journal of Production Economics.

Godinho Filho, M. and Saes, E. V. (2013). From time-based competition (TBC) to quick response manufacturing (QRM): the evolution of research aimed at lead time reduction. The International Journal of Advanced Manufacturing Technology.

Anon. (2026). Bridging compliance and efficiency: leveraging lean manufacturing for ISO 13485 implementation in medical device firms. International Journal for Quality in Health Care, 38(2), mzag065.

Anon. (2022). Industry 4.0 and Lean Six Sigma integration in manufacturing: a literature review, an integrated framework and proposed research perspectives. Quality Management Journal. https://doi.org/10.1080/10686967.2022.2144784

Anon. (2022). A systematic review of the integration of Industry 4.0 with quality-related operational excellence methodologies. Quality Management Journal. https://doi.org/10.1080/10686967.2022.2144783

Anon. (2023). On reliability of reinforcement learning based production scheduling systems: a comparative survey. Journal of Intelligent Manufacturing. https://doi.org/10.1007/s10845-022-01915-2

Anon. (2022). Using process mining to improve productivity in make-to-stock manufacturing. International Journal of Production Research. https://doi.org/10.1080/00207543.2021.1906460

Anon. (2025). Total productive maintenance and Industry 4.0: a literature-based path toward a proposed standardized framework. Applied System Innovation, 8(4), 98.

Moen, R. and Norman, C. Evolution of the PDCA cycle. Working paper, Associates in Process Improvement / The W. Edwards Deming Institute. (Traces the Shewhart cycle of 1939, the Deming Wheel of 1950, Japanese PDCA from 1951, and Deming’s return to PDSA from 1986.)

van der Aalst, W. M. P. (2016). Process Mining: Data Science in Action (2nd ed.). Berlin: Springer.

Books

Imai, M. (1997). Gemba Kaizen: A Commonsense, Low-Cost Approach to Management. New York: McGraw-Hill.

Liker, J. K. (2004). The Toyota Way: 14 Management Principles from the World’s Greatest Manufacturer. New York: McGraw-Hill.

Mann, D. (2014). Creating a Lean Culture: Tools to Sustain Lean Conversions (3rd ed.). Boca Raton, FL: Productivity Press.

Shook, J. (2008). Managing to Learn: Using the A3 Management Process to Solve Problems, Gain Agreement, Mentor and Lead. Cambridge, MA: Lean Enterprise Institute.

Sobek, D. K. II and Smalley, A. (2008). Understanding A3 Thinking: A Critical Component of Toyota’s PDCA Management System. New York: Productivity Press. (Shingo Research and Professional Publication Prize, 2009.)

Goldratt, E. M. and Cox, J. (1992). The Goal: A Process of Ongoing Improvement (2nd rev. ed.). Great Barrington, MA: North River Press.

Rother, M. (2010). Toyota Kata: Managing People for Improvement, Adaptiveness and Superior Results. New York: McGraw-Hill.

Suri, R. (1998). Quick Response Manufacturing: A Companywide Approach to Reducing Lead Times. Portland, OR: Productivity Press.

Standards and regulatory instruments

International Organization for Standardization (2016). ISO 13485:2016 — Medical devices: quality management systems — Requirements for regulatory purposes. Geneva: ISO.

International Organization for Standardization (2019). ISO 14971:2019 — Medical devices: application of risk management to medical devices. Geneva: ISO.

International Organization for Standardization (2019). ISO 11607-1:2019 — Packaging for terminally sterilized medical devices. Geneva: ISO.

International Electrotechnical Commission (2015). IEC 62366-1:2015 — Medical devices: application of usability engineering to medical devices. Geneva: IEC.

International Council for Harmonisation (2009). ICH Q8(R2) — Pharmaceutical Development. Geneva: ICH. (Cited as the reference articulation of the design space construct; not directly applicable to devices.)

International Council for Harmonisation (2008). ICH Q10 — Pharmaceutical Quality System. Geneva: ICH.

US Food and Drug Administration (2024). Quality Management System Regulation, final rule amending 21 CFR Part 820; effective 2 February 2026. Silver Spring, MD: FDA.

US Food and Drug Administration (2026). Inspection of Medical Device Manufacturers, Compliance Program 7382.850. Silver Spring, MD: FDA.

European Union (2017). Regulation (EU) 2017/745 on medical devices (MDR). Official Journal of the European Union.

Note on attribution precision

A small number of entries above are marked “Anon.” or “and colleagues” where the full author list could not be confirmed to a standard adequate for citation at the time of writing. The journal, volume, pages and — where shown — DOI were confirmed. Readers requiring complete author lists should resolve the DOI. No effect size or substantive claim in this document rests on an entry whose bibliographic record could not be confirmed.

Appendix A Evidence Summary Table

Quantified findings referenced in the body, with their sources and the qualifications that should accompany them.

Finding	Source	Qualification
Lean practice bundles explain approximately 23 percent of the variance in operational performance after controlling for industry and context	Shah and Ward (2003), Journal of Operations Management 21(2)	Cross-sectional survey; establishes association, not causation. Also implies roughly 77 percent unexplained.
Six Sigma adoption improves return on assets, driven predominantly by indirect cost reduction; direct cost and asset productivity effects not significant; small sales growth effect	Swink and Jacobs (2012), Journal of Operations Management 30(6)	Matched-sample event study on 200 adopters — among the strongest designs in this literature. Self-selection into adoption is not fully excluded.
Soft lean practices discriminate successful from unsuccessful lean plants; hard practices behave as order qualifiers	Bortolotti, Boscari and Danese (2015), IJPE 160	Survey-based classification of success; culture measured by instrument.
Management support and commitment rank among the highest critical success factors; ranking largely invariant across contingencies	Netland (2016), IJPR 54(8)	432 practitioners, 83 factories, two multinationals. Perceptual measure, not outcome measure.
Theory of Constraints applications report large mean improvements in lead time, cycle time and inventory; no published failures found	Mabin and Balderstone (2003), IJOPM 23(6)	Publication bias explicitly evident. Treat magnitudes as an upper envelope; measurement standardisation across cases is poor.
Zero Defect Manufacturing resolves into four strategies — detect, repair, predict, prevent — across product- and process-oriented implementations	Psarommatis, May, Dreyfus and Kiritsis (2020), IJPR 58(1)	Content analysis of 280 articles, 1987–2018. Taxonomy, not effect estimate.
Digital twin maturity resolves into digital model, digital shadow and digital twin by degree of data-integration automation	Kritzinger et al. (2018), IFAC-PapersOnLine 51(11)	Classification framework. Most industrial deployments marketed as twins are shadows.
Six Sigma effectiveness is explicable through goal theory: specific challenging goals on well-structured tasks with feedback	Linderman, Schroeder, Zaheer and Choo (2003), JOM 21(2)	Theoretical development with propositions. Predicts failure on ill-structured problems.
A Lean Six Sigma project in a medical device firm required nine months, substantially due to validation and regulatory approval activity	McGrane, McDermott, Trubetskaya, Rosa and Sony (2022), Processes 10(11)	Two-case comparison of regulated versus unregulated line transfer. Single firm.
Quality 4.0 defined as traditional quality activities combined with new technologies, expanding rather than replacing quality management scope	Sader, Husti and Daróczi (2022), TQM&BE	Review of 46 articles, 2017–2022. Field still consolidating definitions.
Six Sigma and Industry 4.0 integration amplifies rather than replaces classical method effects	Maia, Lizarelli and Gambi (2024), Benchmarking 31(3)	Systematic review of 59 articles, 2013–2021.
An LSS4.0 framework with a fourteen-step implementation process, evaluated by action research in automotive and mining case settings	Skalli et al. (2025), Production Planning & Control 36(8)	Design science artefact; expert and demonstration evaluation rather than controlled trial.
Fewer than 20 percent of published plan-do-study-act applications reported iterative cycles of change; of those, only about 15 percent reported small-scale testing scaling up with confidence	Taylor, McNicholas, Nicolay, Darzi, Bell and Reed (2014), BMJ Quality & Safety 23(4)	Systematic review of application fidelity, not of efficacy. Healthcare setting; the reporting-quality issue is unlikely to be sector-specific. Implies effectiveness of the method cannot presently be assessed.
5 Whys restricts the analyst to one root cause per causal pathway and to the most distal cause, with no logical basis for treating that cause as the best intervention point	Card (2017), BMJ Quality & Safety 26(8)	Structural critique, not an empirical trial. Author retains the technique’s value as a teaching device for causal reasoning.
Root cause analysis in practice suffers from single-root-cause fixation, variable investigation quality, political context effects, and action plans producing weak risk controls	Peerally, Carr, Waring and Dixon-Woods (2017), BMJ Quality & Safety 26(5)	Analytical review of eight problems with RCA. Healthcare setting; the countermeasure-strength finding transfers directly to device corrective action systems.
Kaizen event outcomes are predicted by organisational and social factors — goal clarity, team composition and autonomy, management support, follow-through — rather than by technical factors; sustainment is the exception, not the default	Farris, Van Aken, Doolen and Worley (2009), IJPE 117(1); Glover, Farris, Van Aken and Doolen (2011), IJPE 132(2)	Field studies of 51 events in six organisations and 65 events in eight organisations respectively. Among the strongest empirical work on any daily-management practice.
Cross-training is a design space with distinct architectures and coordination policies; the mechanism must be specified before the architecture is chosen	Hopp and Van Oyen (2004), IIE Transactions 36(10)	Analytical framework with queueing and simulation support; field validation of specific architectures is thinner.
The Shainin System isolates a dominant cause by progressive elimination; efficient where a dominant cause exists, unsound where it does not	Steiner, MacKay and Ramberg (2008), Quality Engineering 20(1)	Peer-reviewed overview and critical assessment. Authors note much of the system was previously not discussed in peer-reviewed literature.
Applying visual management principles to shop-floor communication boards improved team leaders’ ability to engage teams in problem solving	Bateman, Philp and Warrender (2016), IJPR	Two-year longitudinal single-company study. Rare longitudinal evidence on a daily-management artefact.

Appendix B Maturity Assessment Rubric

A short instrument for locating a plant on the Table 6.1 scale. Each item is scored 0 (absent), 1 (partial) or 2 (established) on the basis of observable evidence rather than of documented intent. A level is achieved when every item at that level and below scores 2.

Level	Assessment item	Evidence required
2	Loss is decomposed in currency terms by category and by mode	A current decomposition reconciling to the financial ledger
2	Measurement system adequacy is quantified for every critical measurement	Attribute agreement and gauge study results with dates, including inspection false-discovery rate
2	Baseline capability exists for every critical-to-quality characteristic	Capability studies with a documented rational subgrouping scheme
2	Standardised work exists, built with operators, and is separated from the controlled work instruction	Posted standard work sheets with takt, sequence and standard WIP; cross-reference to the controlled instruction
2	Tiered accountability runs daily and time-to-detection is measured	Meeting cadence with escalation records; trended interval from abnormality to raise
2	Skills matrix uses observable competence criteria and is reviewed operationally	Assessment criteria per level; coverage gaps with owners and dates; evidence of a staffing decision made from it
3	Problem-solving method is selected by problem structure, not by habit	Method selection recorded against Table 4.1 for the last twenty investigations
3	Root causes are verified experimentally rather than asserted	Evidence that the cause was shown to turn the effect on and off
3	Countermeasures are classified by strength and the profile is managed	Distribution of the last fifty corrective actions across the Table 4.2 hierarchy, trended
3	Corrective action effectiveness criteria are defined before implementation	Pre-defined statistical criterion, observation period and sample size in the action record
4	A3 practice is mentored and iterated	Median draft count per completed A3 above two; named mentors with capacity
3	Improvement capacity is committed and delivery is tracked against commitment	Hours committed and hours delivered, by month, for the last four quarters
3	Projects are selected from the loss decomposition and the risk file	Project charters cross-referenced to risk file entries and to quantified loss
3	Control mechanisms are verified at 180 days with finance validation of benefit	Sustainment records and the ratio of validated to claimed benefit
3	Constraint is identified and exploitation precedes elevation	Constraint analysis and a documented exploitation review preceding the last capital request
4	Line-level personnel run structured experiments as routine	Experiment records with prior predictions, by learner, over the last quarter
4	Design spaces exist for critical validated processes	Validation reports whose validated object is a region, not a point setting
4	Quality-system processes are measured and improved as processes	Process mining or equivalent analysis of deviation and change control with resulting interventions
4	Change-control throughput is measured and managed as a constraint	Median request-to-implementation time and queue length, trended
5	Defect modes are addressed by prediction and prevention rather than detection	Defect taxonomy with strategy assignment and a trend in the prevention share
5	Maintenance is condition-driven within validated envelopes	Condition-based scheduling records with retained validated maximum intervals
5	Experimentation is substantially in-silico with validated models	Model validation records with stated regions of validity, and physical-run avoidance data
5	Model governance is operating	Version control, performance monitoring against qualification baseline, retraining triggers, rollback path

Appendix C Glossary

Term	Definition as used in this document
A3 process	A one-page structure — background, current condition, target, cause analysis, countermeasures, plan, follow-up — used as a coaching mechanism through iterated drafts with a mentor, not as a reporting format.
Attribute agreement analysis	A measurement system study for categorical (pass/fail) measurement, quantifying agreement of the measurement system with a known truth panel and with itself on repeat measurement. The correct study type for automated optical inspection.
Catchball	The negotiation protocol in Hoshin Kanri by which each organisational level proposes the means by which it will contribute to the objectives of the level above, and receives acceptance or modification.
Change-control tax	Term used here for the additional effort per process change imposed by a regulated quality system: risk assessment, impact analysis, qualification, documentation and, where applicable, notification.
Coefficient of variation (failure interval)	Standard deviation divided by mean of the time between failures. A low value implies time-based maintenance is near-optimal; a high value implies condition-based maintenance has value.
Daily management system	The L0 substrate: standardised work, leader standard work, visual controls, tiered accountability and a working problem-solving method. Determines the interval from abnormality to detection and from detection to resolution.
Design space	The multidimensional combination of material attributes and process parameters demonstrated to provide assurance of quality. Movement within an established design space is not, under the pharmaceutical framework in which the term originates, a change.
Digital model / shadow / twin	The Kritzinger et al. (2018) taxonomy by degree of data-integration automation: no automated exchange; automated one-way physical-to-digital; automated bidirectional.
Drum-buffer-rope	The Theory of Constraints scheduling mechanism: constraint schedule (drum), time buffer protecting the constraint (buffer), and material release tied to constraint consumption (rope).
Escape point	In 8D, the point in the control system at which the defect should have been detected but was not. A second causal question, independent of why the defect was created.
Every-part-every-interval (EPEI)	The interval within which a process cycles through its full product complement. A direct function of changeover time and batch policy.
Gemba / genchi genbutsu	The actual place where value is created; and the practice of going to see it directly rather than managing through reports, on the argument that aggregation removes precisely the variance and anomaly that carry diagnostic information.
Leader standard work	Documented recurring activities, at defined frequency and location with defined outputs, for each level of management. Standardised proportion falls with seniority — high at team leader, low at plant manager.
Manufacturing critical-path time	The QRM measure of lead time: typical calendar time from order receipt to first delivery, assuming the order starts from nothing.
Order qualifier / order winner	Terry Hill’s distinction between attributes required to compete and attributes that determine who wins. Applied by Bortolotti et al. (2015) to hard and soft lean practices respectively.
Overall equipment effectiveness	Availability × performance rate × quality rate, with losses decomposed into breakdowns, setup and adjustment, minor stoppages and idling, reduced speed, start-up rejects and process defects.
POLCA	Paired-cell overlapping loops of cards with authorisation. A hybrid push-pull control mechanism using card loops between cell pairs rather than per part number, designed for high-variety environments.
Quantity at risk	The volume of product produced between defect creation and defect detection. The principal driver of the economic value of detection-latency reduction.
PDCA / PDSA	The improvement cycle. Deming objected to “check”, which connotes inspection, and returned to “study”, which denotes learning from the comparison of result against prediction. The distinction marks the difference between an experimental discipline and a project checklist.
Rational subgrouping	The scheme by which observations are grouped for control charting so that within-subgroup variation represents common cause only. Incorrect subgrouping is a leading cause of uninformative control charts.
Soft lean practices	Small-group problem solving, training, supplier and customer involvement, and continuous-improvement leadership — distinguished from hard practices such as pull, setup reduction and statistical process control.
Standardised work	Takt time, work sequence and standard work-in-process for a task: the documented current best-known method, and therefore the baseline against which abnormality becomes visible.
Statistical engineering	The discipline, in the sense of Hoerl and Snee, of using statistical concepts and methods in integrated sequences to attack large, unstructured, high-consequence problems.
Zero Defect Manufacturing	A control architecture organised around four strategies — detection, repair, prediction and prevention — applied in product-oriented and process-oriented forms.