If I could start over — clean sheet, no legacy systems, no inherited procedures, no "we have always done it this way" — what quality system would I build? After twenty years across Airbus, SNOP, WITTE Automotive, and a dozen consulting engagements, I have a clear answer. It looks nothing like the quality systems most companies run.
It is simpler, harder, and more effective. Here it is.
Principle 1: Build for behaviour, not documentation
The first thing I would do is resist the urge to document everything. Most quality systems are over-documented because documentation feels like control. It is not. Documentation is evidence of control, not control itself. Control comes from behaviour — from people doing the right things consistently because they understand why those things matter.
My clean-sheet system would have one-tenth the procedures of a typical ISO 9001 system. Each procedure would fit on two pages and be written in plain language. The rest of the system would live in the process itself — in visual management at the point of use, in standard work embedded in the equipment, in mistake-proofing that makes the wrong action difficult or impossible.
The quality manual would be twelve pages. The procedure for nonconformance management would be one flowchart. The training system would be competence-based, not attendance-based. The audit programme would be risk-based, focusing on processes that matter, not on blanket coverage that satisfies a clause.
The quality system should be invisible to the operator and visible to the leader. If your operator is thinking about the quality system, they are not thinking about the product.
Principle 2: Measure what predicts, not what reports
Every quality system I have inherited measures lagging indicators — defect rates, customer complaints, audit findings, nonconformance counts. These metrics tell you what happened. They do not tell you what will happen.
My clean-sheet system would lead with predictive metrics: process capability trends, SPC out-of-control signals, equipment condition data, supplier leading indicators, competence assessment scores, maintenance compliance rates. These metrics tell you what is about to happen, giving you time to act before the defect occurs.
The dashboard would have five metrics, not fifty. Each metric would have a trend line, not a traffic light. And every deviation would trigger a conversation, not a colour change.
Principle 3: Make prevention the default
Most quality systems are reactive — they respond to defects after they occur. The prevention activities — PFMEA, control planning, process design, mistake-proofing — exist as documents, not as daily practice. They are created at launch and forgotten at serial production.
My clean-sheet system would make prevention continuous. The PFMEA would be a living document, reviewed at every engineering change, every equipment modification, every supplier change. The control plan would be updated whenever the process capability shifted. Mistake-proofing would be a design requirement, not a corrective action — every new process would be designed with failure prevention built in, and the engineering team would not be allowed to release a process without demonstrating that failure modes have been addressed at the source.
This is hard. It requires engineering discipline that most organisations do not enforce. But it is the single highest-leverage investment in quality, and it is the difference between a system that prevents defects and a system that detects them.
Principle 4: Give authority to the point of detection
In my clean-sheet system, every operator would have andon authority — the right and the obligation to stop the process when they detect an anomaly. No approval needed. No permission required. The operator stops, the team leader responds within minutes, and the issue is resolved or escalated immediately.
This principle extends beyond the shop floor. Every quality engineer would have the authority to hold nonconforming product without needing managerial approval. Every supplier quality engineer would have the authority to escalate a supplier issue to the procurement director without going through three layers of management. Speed of response is a quality metric, and hierarchical approval is the enemy of speed.
Principle 5: Integrate, do not isolate
The quality system would not be a separate system. It would be embedded in the operational system. Quality metrics would appear on the same dashboard as production metrics, safety metrics, and cost metrics — not on a separate quality board reviewed in a separate quality meeting. The message: quality is not a function. It is a dimension of every function.
The quality team would not sit in a separate office. They would be physically located with the operations team. Their daily standup would be the operations standup. Their improvement projects would be operations improvement projects. The boundary between quality and operations is artificial, and in my clean-sheet system, it would not exist.
Principle 6: Make the cost of quality visible
The financial impact of quality — both the cost of good quality and the cost of poor quality — would be measured monthly and presented in the same format as the financial reports. Not in a separate quality cost report that nobody reads. In the P&L review, alongside revenue, margin, and operating cost.
Scrap, rework, warranty, customer credits, expedited freight, quality team overtime — all of it would be visible, categorised, and trended. The executive team would see the cost of quality every month, and the quality investment proposals would be evaluated as business cases with expected returns, not as cost centre requests.
Principle 7: Hire for thinking, train for skills
The quality team I would build is small — smaller than most companies would expect. Five people for a 500-employee plant: a quality manager, two quality engineers, a supplier quality engineer, and a continuous improvement lead. But every one of them would be a systems thinker, not a compliance manager.
I would hire for analytical thinking, communication skills, and the courage to speak truth to power. I can teach statistical methods. I can teach 8D methodology. I can teach audit technique. I cannot teach someone to think systemically or to stand their ground when the plant manager says "ship it." I hire for the things I cannot teach and train for the rest.
The system I would not build
Equally important is what I would leave out. No quality manual that restates the standard. No procedure library that nobody reads. No traffic-light dashboards that trigger defensive behaviour. No annual management review that is actually a status meeting. No corrective action system that rewards closure over effectiveness. No supplier scorecard that measures compliance over partnership. No training records that confuse attendance with competence. No risk register that is maintained for the audit instead of for the business.
Every tool I would leave out is a tool that exists in every quality system I have inherited. They are there because someone, at some point, created them to satisfy a requirement. And then they became permanent. They take time, energy, and attention away from the things that actually improve quality, and they create the illusion of a functioning quality system while draining the life from it.
If I started over tomorrow, I would build a quality system that is lean, behavioural, preventive, integrated, and honest. It would not impress an auditor with its documentation. It would impress a customer with its results. And in the end, that is the only impression that matters.
