💡 Concept page · Deming · System thinking in management practice

Deming’s 14 Points — plain English, with Joiner levels and NHS examples

W. Edwards Deming identified most Level 3 system interventions decades before Joiner gave them a classification. The 14 Points are not a management checklist — they are a structural description of what a system that produces quality looks like, and what stands in the way of building one. Most organisations implement none of them.

Who was Deming and why does he matter?

W. Edwards Deming was an American statistician and management theorist who transformed Japanese manufacturing after World War II. Toyota, Sony, and the companies that became synonymous with quality in the 1970s and 1980s built their operating systems on his ideas. His influence on Japanese industry was so significant that Japan’s most prestigious quality prize — the Deming Prize — has been awarded since 1951.

Western management largely ignored him until the 1980s, when it became impossible to deny that Japanese manufacturers were consistently outcompeting their American and European rivals. Deming spent the last decades of his life teaching American companies what he had spent decades teaching Japan. He published Out of the Crisis in 1982, at the age of 82. It remains the most important book on quality management ever written.

His relevance to this site is direct. Deming was a statistician first. His entire framework is built on the idea that you cannot improve what you cannot measure honestly — and that most measurement in organisations is dishonest, either by design (targets that incentivise gaming) or by method (charts that cannot distinguish genuine change from noise). Bootstrap CUSUM is the tool that makes the Study phase of his PDSA cycle honest. The 14 Points are the management system that gives Level 3 interventions their practical form.

📉 The central insight — Deming’s 94% rule
94%

Deming estimated that 94% of problems in any organisation are caused by the system — the structures, processes, policies, and management practices that determine how work gets done. Only 6% are caused by individual workers. This is the foundation of every one of the 14 Points. If 94% of problems are system problems, then 94% of improvement effort should be aimed at the system. Most organisations do the opposite: they react to symptoms, train individuals, set targets, and apply pressure — all Level 1 and Level 2 interventions aimed at the 6% while the 94% continues unchanged.

Blame the system, not the people. If 94% of problems are caused by the system, then 94% of blame directed at individuals is misdirected. The worker is doing their best within a system that is producing the problem. The job of management is to change the system — not to find better people to put inside a broken one.
⚠️ The connection to Joiner

Joiner noted explicitly that most Level 3 system fixes are already described in Deming’s 14 Points. Deming identified them decades earlier — but without the level classification that makes clear why they work when lower-level interventions do not. Read the 14 Points through Joiner’s lens and the mechanism becomes precise: most of the Points are Level 3 or Level 3 Deep interventions. The ones that are not — Points 10 and 11 — are Deming’s explicit rejection of Level 1.

The 14 Points — plain English with Joiner levels

Each Point is explained in plain English, classified by Joiner’s level, and illustrated with an example from the NHS or public sector data on this site where available.

1
Create constancy of purpose for improvement
Level 3 Deep

The organisation must have a clear, stable, long-term purpose — one that does not change with every new leader, every new political cycle, or every new performance framework. Without constancy of purpose, every structural improvement is eventually dismantled when the next reorganisation arrives.

Why Level 3 Deep: This is Meadows’ paradigm level — the mental model from which the system’s goals, rules, and structures flow. You cannot achieve it by changing a process or writing a policy. It requires the organisation to genuinely commit to improvement as its primary purpose rather than as a periodic initiative.

NHS example: the A&E four-hour target has been the subject of fourteen different policy initiatives in fifteen years. Bootstrap CUSUM finds four stages of structural decline. No constancy of purpose — no sustained structural improvement.
2
Adopt the new philosophy
Level 3 Deep

Reject the acceptance of defects, delays, and poor quality as inevitable. The “new philosophy” is that quality is not an inspection problem — it is a design problem. The system must be designed to produce quality, not to detect failure after it has occurred.

Why Level 3 Deep: This is a fundamental reorientation of what the organisation is for. It cannot be achieved by a training programme or a new process. It requires management to change what they believe about where quality comes from.

Healthcare example: wrong-route medication administration was declared “wholly preventable” in 2011. The engineering solution (NRFit connectors) existed. Adopting it would have required accepting that the problem was a design failure, not a training failure. The training-based approach continued. Bootstrap CUSUM finds a flat process at 17.5 events per year.
3
Cease dependence on inspection to achieve quality
Level 3

Inspection at the end of a process finds defects after they have been created. It does not prevent them. Build quality into the process so that inspection becomes unnecessary. “Build it right first time” is not a slogan — it is the result of designing a system that cannot easily produce defects.

Why Level 3: Stopping dependence on inspection requires changing the process design, not just the inspection protocol. It is an engineering control in Joiner and COMAH terms — the hazard is addressed at source rather than detected downstream.

Manufacturing example: in pharmaceutical-grade chemical production, inspection of the final product does not improve it. The specification is met or not met. The improvement is in the process stability — detected by CUSUM on the process parameters, not by end-product testing alone.
4
End the practice of awarding business on price alone
Level 3

Lowest-cost procurement produces lowest-quality input. Move toward single suppliers for important items, building long-term relationships based on quality and total cost rather than unit price. The purchasing system produces the quality problem — changing the system changes the output.

Why Level 3: This is a structural change to the procurement system, not a request for buyers to make better decisions within the existing system. It requires changing the rules and incentives that govern purchasing, not just the preferences of individual buyers.

NHS example: NHS procurement frameworks that award contracts primarily on price produce lowest-cost suppliers who then cut corners on quality and reliability. The cost of failure far exceeds the saving on unit price — but the saving is visible on the procurement dashboard and the cost of failure appears elsewhere in the system.
5
Improve constantly and forever
Level 3

Improvement is not a project with a start and an end date. It is a permanent structural discipline built into every process. PDSA as a continuous cycle — not as a response to a crisis or a regulatory requirement. The system is designed to improve itself.

Why Level 3: Embedding continuous improvement requires changing the structure of work so that improvement activity is part of every role — not an additional task assigned to a separate team. It is a system design change, not a behavioural request.

Bootstrap CUSUM connection: a system that improves constantly will show a progressive downward staircase of step changes on the Bootstrap CUSUM chart — each cycle producing a new confirmed lower mean. A system that improves episodically and then reverts will show step changes followed by reversion. The chart shape tells you whether improvement is structural or cyclical.
6
Institute training on the job
Level 2

Workers cannot do good work if they do not know what good work looks like or how to achieve it. Training is necessary. But training alone is Level 2 — it changes what people do within the existing system without changing the system that determines what is possible.

Why Level 2: Deming and Joiner both note that training is necessary but insufficient without system change. A well-trained worker in a poorly designed system will still produce poor-quality output — because the system determines the outcome, not the individual’s effort or knowledge.

NHS example: dementia awareness training for GPs is a Level 2 intervention. It may improve individual behaviour within the existing pathway. It does not address the constraint — memory clinic capacity — that limits the pathway’s throughput regardless of how well-trained the referring GP is.
7
Institute leadership
Level 3

The job of management is not to supervise and control workers — it is to improve the system that workers operate in. Leadership in Deming’s sense means helping people do better work by understanding and changing the system, not by monitoring performance metrics and applying pressure when targets are missed.

Why Level 3: This requires changing the role of management structurally — from directing outputs to improving systems. It is a fundamental change to what managers are responsible for and how their performance is evaluated. It cannot be achieved by a leadership training course.

Organisation example: a manager who responds to every missed target by pressuring the team to work harder is a Level 1 manager in a Level 3 role. The system continues to produce the same results. A manager who responds by asking what in the system produced the miss — and changes that — is applying Deming’s Point 7.
8
Drive out fear
Level 3 Deep

Fear of reporting problems, fear of raising concerns, fear of missing targets — all produce the same result: people hide what is actually happening. Bad data flows upward. Good decisions cannot be made. The organisation optimises for the appearance of performance rather than its reality.

Why Level 3 Deep: Fear is a system condition produced by structures that blame individuals for system failures. It requires structural change to how performance is evaluated, how problems are reported, and whether the messenger is punished. It cannot be addressed by asking people to be braver.

NHS example: the Mid Staffordshire inquiry documented how fear of reporting poor performance allowed serious failures to persist for years. The cultural conditions that produced that fear were structural — a board focused on financial targets, a regulatory environment that penalised admission of failure, and a management culture that treated concerns as threats rather than data.
9
Break down barriers between departments
Level 3

Departments that optimise for their own performance measures produce system-level failure. Quality problems, handover failures, and coordination costs are almost always the result of structural siloes — not of individuals who do not want to cooperate. Breaking down the barriers requires changing the structure, not exhorting people to collaborate more.

Why Level 3: Siloed departments are a structural design. Changing them requires changing reporting lines, shared measures, physical co-location, or integrated information systems — structural interventions, not cultural ones.

NHS example: the dementia pathway fails at the boundary between primary care (GP referral), secondary care (memory clinic diagnosis), and social care (post-diagnostic support). Each part is optimised for its own metrics. The patient journey through the boundary is nobody’s explicit responsibility.
10
Eliminate slogans, exhortations, and targets for the workforce
Anti-Level-1

Slogans that ask workers to try harder, work smarter, or achieve zero defects do nothing to change the system that produces defects. They shift responsibility from the system — where Deming says 94% of problems originate — to the individual worker. Deming argues this is not just ineffective but actively harmful: it damages morale, produces adversarial relationships, and obscures where the real problem lies.

Why Anti-Level-1: Slogans are Level 1 interventions — they react to outcomes without changing the system that produces them. Deming’s Point 10 is his explicit rejection of this approach. The system produces the result. Change the system.

NHS example: “zero harm” campaigns, patient safety slogans, and targets for never events fall into this category when they are not accompanied by system-level changes. Bootstrap CUSUM on NHS never events data finds a flat process — the slogans and targets have not moved the line.
11
Eliminate numerical quotas and management by objectives
Anti-Level-1

Numerical quotas without a method for achieving them are goals without a system change. They produce exactly what Deming predicted: gaming, distortion, and measurement that optimises the number rather than the outcome it was intended to represent. The quota replaces the aim. People hit the number by changing how it is counted, not by improving what it is supposed to measure.

Why Anti-Level-1: A target is a Level 1 intervention. It reacts to an output without changing the system that produced it. Bootstrap CUSUM applied to any metric governed primarily by a numerical target will typically find temporary movement followed by reversion — or, worse, a persistent change in the metric that reflects changed coding rather than changed outcomes.

NHS example: the dementia 66.7% diagnosis target was set in 2012. By 2019 England briefly exceeded it. Bootstrap CUSUM finds one stage, mean below target, no sustained structural improvement. The target measured the number of diagnoses recorded. It did not change the system that constrained how many diagnoses could be made or what happened after diagnosis.
12
Remove barriers that rob workers of pride in workmanship
Level 3

Workers who cannot do good work because the system prevents it — inadequate tools, unclear standards, performance evaluations that measure the wrong things, managers who blame them for system failures — lose the motivation to improve. The system has taken away their ability to take pride in their work.

Why Level 3: Pride in work is destroyed by system conditions, not by individual attitudes. Restoring it requires structural changes to how work is designed, how performance is evaluated, and how problems are attributed. It cannot be achieved by asking workers to take more pride.

NHS example: clinicians who see patients readmitted because the discharge process is inadequate — and who cannot change the discharge process — experience exactly this. The system prevents them from doing the work they know is right. The result is disengagement and resignation, not improved performance.
13
Institute a vigorous programme of education and self-improvement
Level 3

Not training for the current job — education that develops people’s capacity to understand and improve the system itself. Statistical thinking, systems thinking, improvement methodology. The organisation that educates its people to understand the system they work in builds its own internal improvement capability rather than depending on external consultants or periodic initiatives.

Why Level 3: Building internal improvement capability requires a structural commitment — time protected for education, career paths that reward improvement knowledge, and management that uses what people learn. It is a system design change, not a training budget line.

14
Put everyone to work to accomplish the transformation
Level 3 Deep

The transformation to a quality organisation is not a project managed by a quality department. It requires top management to lead it, commit to it permanently, and involve everyone in it. Deming’s Point 14 is the application of Point 1’s constancy of purpose to the transformation itself: the commitment must be genuine, visible, and sustained — not announced and then quietly abandoned when the next crisis arrives.

Why Level 3 Deep: This requires the entire organisation’s mental model to change — from “improvement is something the quality team does” to “improvement is everyone’s job in every role every day.” It is the definition of constancy of purpose in action.

The connection to Bootstrap CUSUM: if Point 14 is genuinely in place, the Bootstrap CUSUM chart for every key metric will show a progressive downward staircase over years — each PDSA cycle producing a confirmed new lower mean. That is what a transforming organisation looks like in data.

Points 10 and 11 in depth — why targets without method fail

Points 10 and 11 deserve special attention because they are the most commonly violated and the most directly relevant to the Bootstrap CUSUM findings on this site. Together they represent Deming’s explicit, forceful rejection of management by targets.

The mechanism of target failure. A numerical target set without a method for achieving it does four things. First, it shifts attention from the system to the metric. Second, it creates pressure to hit the number through any available means — including changing how the number is counted. Third, it obscures the constraint that is actually limiting performance — because people are focused on the metric, not the system. Fourth, it relieves the pressure for structural change by producing the appearance of improvement when the number temporarily moves — even if nothing in the system has changed. Bootstrap CUSUM strips away the appearance and shows the underlying structure. That is why the NHS data on this site is so consistent: targets without system change produce no confirmed step change.
“A goal without a method is nonsense. Management by numerical goal is an attempt to manage without knowledge of what to do.” — W. Edwards Deming

The dementia 66.7% diagnosis target is the clearest illustration on this site. Set in 2012. Briefly exceeded in 2019. Bootstrap CUSUM finds one stage, mean below target, no confirmed structural improvement. The target measured the number of diagnoses recorded. It did not change the constraint — memory clinic capacity — that limited how many diagnoses could be made. It did not change what happened to patients after diagnosis. It did not change the reinforcing loops that kept the pathway reactive rather than proactive. It was a numerical quota without a method. Deming predicted exactly this result eighty years ago.

Quick reference — all 14 Points by Joiner level

Point Title Joiner level
1Constancy of purposeLevel 3 Deep
2Adopt the new philosophyLevel 3 Deep
3Cease dependence on inspectionLevel 3
4End lowest-cost supplier policyLevel 3
5Improve constantlyLevel 3
6Institute trainingLevel 2
7Institute leadershipLevel 3
8Drive out fearLevel 3 Deep
9Break down barriersLevel 3
10Eliminate slogans and targetsAnti-Level-1
11Eliminate numerical quotasAnti-Level-1
12Remove barriers to prideLevel 3
13Institute educationLevel 3
14The transformation is everyone’s jobLevel 3 Deep
The pattern is stark. Ten of the 14 Points are Level 3 or Level 3 Deep interventions. One is Level 2. Two are explicit rejections of Level 1. Not one Point is a Level 1 firefighting intervention. Deming is not describing a list of good practices — he is describing a system redesign. Most organisations implement Point 6 (training) and ignore the rest. Then they wonder why nothing changes.

Deming and Bootstrap CUSUM — the measurement connection

Deming invented the PDSA cycle — Plan, Do, Study, Act. He was insistent that the Study phase is the most important and the most neglected. Most organisations plan and do, then declare success or failure without genuinely studying the evidence. The study requires honest measurement — a tool that can distinguish genuine structural change from noise.

Bootstrap CUSUM is that tool. It sits in the Study phase of every PDSA cycle on this site. When a policy is implemented, a target is set, or a structural intervention is made, Bootstrap CUSUM answers the question Deming demanded: has the process actually changed, or are we seeing variation around an unchanged mean?

Deming’s tampering trap and Bootstrap CUSUM. Deming identified tampering as one of the most common causes of persistent failure: intervening again before the first intervention has had time to produce its effect. Bootstrap CUSUM prevents tampering by making the evidence explicit — it will not confirm a structural change until the statistical evidence warrants it. A confirmed step change on the Bootstrap CUSUM chart is Deming’s Study phase completed honestly. Run Bootstrap CUSUM on your own improvement data ›

Where this fits — and where to go next

🎯 Our mission

“Give people better tools and a better mental model to ask better questions about whether things have improved through structural change — and to understand why, when they haven’t.”

▶ Run Bootstrap CUSUM on your data Joiner levels of fix ›