The Knowledge System Podcast

A leader can make a process worse while trying very hard to improve it. That is the danger of tampering. When every disappointing result triggers a new rule, a new explanation, or a new adjustment, management may be reacting to routine variation as if something unusual happened.

The result is more noise, more frustration, and less learning. The better question is not, What changed yesterday? The better question is, Do we know what kind of variation we are seeing?

The temptation to chase the last number

W. Edwards Deming used a simple funnel experiment to make this problem visible. Imagine dropping a marble through a funnel toward a target. Even if the funnel stays in the same position, the marble will not land in exactly the same place each time.

The natural impulse is to move the funnel after each miss, trying to compensate for the last result. That feels sensible. It feels active. It feels like control. But if the process is already stable, repeated adjustment can spread the results farther from the target.

The act that feels like control becomes a source of instability. Management can fall into the same pattern whenever it treats the latest result as a command.

A dashboard turns red. A customer complains. A weekly number dips. Someone asks for an explanation, and the organization rushes to change the work. Sometimes that response is necessary. A real special cause deserves attention. But when the process is stable, the better work is to improve the system, not chase each point.

That is the problem ClearStep, a mid-sized B2B software company, faced when its support leaders tried to improve response time by changing the process after every bad day.

The support dashboard that would not settle down

ClearStep sold project management software to manufacturers. Its support team handled setup questions, bug reports, billing issues, and urgent support calls. The team was capable, but its work arrived unevenly.

Rina, ClearStep’s head of customer support, watched one number more than any other: median first response time. When it rose, customers complained. When it fell, the executive team relaxed.

Monday morning, the dashboard looked bad. Response time had jumped from twenty-three minutes to thirty-seven. Rina opened the team meeting with a decision already forming.

“We need a new rule. For the rest of the week, no one works on follow-up tickets until the new queue is under control.”

Marcus, the operations analyst who helped the support team study workflow data, hesitated. He had been plotting daily response time for the past six months.

“I know thirty-seven minutes looks bad,” Marcus said. “But it is still inside the range we have seen before.”

“Customers do not care about ranges,” Rina said. “They care that we were slow.”

“Agreed,” Marcus said. “But if we change the rule every time the number moves, we may be adding variation ourselves.”

That was not what Rina wanted to hear. She was trying to be responsive, not careless. The team had already changed the escalation rule twice that month. One week, senior agents took every urgent ticket first. The next week, new tickets came first.

By Thursday, response time improved, but reopenings were up. Customers got quick replies that did not resolve the issue. The team was moving faster and learning less.

Deming named the trap plainly: “Mistake 1. To react to an outcome as if it came from a special cause, when actually it came from common causes of variation.”

Mistake 1. To react to an outcome as if it came from a special cause, when actually it came from common causes of variation.— W. Edwards Deming

Rina asked Marcus to show the chart again. The bad Monday was unpleasant, but it was not outside the usual pattern. The system had been predictable for months. Response time bounced within a wide band because of uneven ticket routing, inconsistent urgency definitions, and too few agents trained on integration issues.

“So doing nothing is the answer?” Rina asked.

“No,” Marcus said. “Studying the system before we change the rules is the answer.”

“Then what do we change?”

“Not the queue every morning. We change the conditions that keep creating these wide swings.”

That distinction changed the conversation. ClearStep still investigated real signals: outages, product releases, unusual customer spikes. But it stopped rewriting queue rules after ordinary variation. Rina’s team clarified urgency definitions, cross-trained agents on integration questions, and reviewed blocked tickets each day to remove causes of delay.

The solution was not inaction. It was action aimed at the system.

Why we keep treating noise like a signal

We drift into tampering because the pressure to respond is real. A leader sees a bad number and feels responsible for it. A customer is waiting. A team is anxious. An executive wants an explanation. In that moment, studying variation can sound like delay.

But the demand for an explanation can create its own distortion. If every up and down requires a story, people will supply stories. Some will be true. Some will be guesses. Some will be shaped by what seems safest to say. The organization may become better at explaining variation than reducing it.

Deming’s warning is uncomfortable because it challenges a common picture of leadership. We often equate visible reaction with accountability. We expect the manager to change something, tighten something, or call someone into the room. But if the latest point came from common causes, the visible reaction may make the system harder to understand.

Deming put the problem sharply: “They were tampering with a stable system, making things worse.”

This is not only an internal efficiency problem. A company that keeps changing priorities teaches employees to protect themselves from the latest swing. It teaches customers to expect inconsistency. Over time, reliability becomes harder to deliver, and trust becomes harder to earn.

The management habit that looks decisive in the moment can quietly weaken the system that customers experience.

What leaders can do instead

Before choosing a response, leaders need a way to separate movement that calls for investigation from movement that calls for system improvement. Deming was not arguing for passivity. He was arguing for action that fits the evidence.

* Observe the system before reacting. Do not treat the latest point as a command. Look at performance over time and ask whether the process is showing a real signal or behaving as it has behaved before.

* Separate urgency from interpretation. A customer problem may need immediate care, but that does not mean the process itself needs a new rule. Serve the customer, then study what the result means.

* Ask better management questions. Instead of asking, “Who caused yesterday’s miss?” ask, “What conditions keep producing this range of results?” The second question moves attention from blame to capability.

* Improve the sources of variation. Work on definitions, training, handoffs, equipment, priorities, and decision rules. These are less dramatic than a new order from management, but they are usually closer to the real causes.

* Build reliability as a management advantage. When a company reduces unnecessary variation, people can plan, customers can trust the service, and leaders can learn faster. That consistency is hard to copy because it comes from the way the system is managed.

The key is to avoid confusing energy with improvement. Activity can make management feel involved while leaving the process worse than before.

Better leadership starts by seeing variation

Tampering is tempting because action feels responsible. But leadership is not measured by how quickly a manager changes something after a bad number. It is measured by whether the action fits the evidence.

Some results call for investigation. Some call for patience and system improvement. The skill is knowing the difference. When leaders stop chasing noise, they create the conditions for steadier work, clearer learning, and better performance tomorrow.

That is the practical promi…

Most people do not begin meaningful work hoping to do the minimum. They want to contribute, solve problems, serve people well, and take pride in what they do. Yet many organizations manage as if motivation must be manufactured from the outside through rankings, bonuses, contests, pressure, or fear.

W. Edwards Deming saw a deeper problem: management can either protect the human desire to learn and contribute, or quietly damage it. Quality depends on judgment, cooperation, and learning. Those cannot be forced into existence.

The harder question behind performance

It is easy to assume that poor performance means people need more pressure. When results disappoint, leaders often reach for sharper targets, clearer rankings, stronger incentives, or more visible accountability. These methods feel practical because they create attention quickly.

But attention is not the same as improvement. People can pay attention to a score while the work gets worse. They can learn how to look good on a dashboard while customers experience delay, confusion, or uneven service.

Deming placed motivation inside the psychology element of his System of Profound Knowledge. His warning was not that pay, goals, or recognition have no effect. It was that leaders must understand what these devices do to people, especially when they replace purpose, learning, and cooperation.

He stated the danger plainly: “Extrinsic motivation in the extreme crushes intrinsic motivation.”

Extrinsic motivation in the extreme crushes intrinsic motivation.— W. Edwards Deming

Northstar Clinics shows how easily a reasonable performance idea can become a barrier to better work.

The score was not the same as the work

Northstar Clinics operated nine outpatient clinics. Wait times were uneven. Access was slipping. Turnover was rising. Elena, the operations leader, wanted a plan with force to change behavior.

She came to a leadership meeting with a dashboard proposal. Each clinic would receive a monthly productivity score. The top clinic would be recognized; the bottom clinic would submit a plan.

Elena explained the idea directly.

“We need people to know this matters. If we recognize the top performers, the others will have a reason to catch up.”

Marcus studied the draft dashboard. He understood why Elena wanted accountability, but something about the design bothered him.

“Maybe. But what if the score changes what people pay attention to?”

Elena pushed back. “They should pay attention to access, callbacks, and visit flow. That is the point.”

“Or they may pay attention to looking good on the dashboard,” Marcus said. “A clinic can lift the score and still make the work worse.”

That was the uncomfortable turn. Elena wanted focus. Marcus was asking whether the proposed system would improve the work or merely change behavior around the measurement.

“Then what are you suggesting? We cannot just ask everyone to care more.”

Marcus answered quietly.

“I do not think caring is the problem. I think the system is wearing people down.”

The room went still. The issue was no longer whether the dashboard was clear enough. The issue was whether management understood the conditions under which people were working.

The team began studying the clinics instead of ranking them. One served more complex patients. Another had lost two exam rooms to equipment problems. A third had nurses covering refills, triage, and insurance paperwork. These differences were not excuses. They were part of the system producing the results.

Elena visited one clinic the following week. She watched a medical assistant search for a working blood pressure cuff while a physician waited for misrouted lab results. No one looked indifferent. They looked worn down by repeated obstacles.

Later, Elena asked a nurse what would help.

“If you could change one thing about the system, what would it be?”

Marcus added, “Take your time. This is not a performance review.”

That sentence mattered. People were used to explaining bad numbers, not naming barriers without fear.

Deming connected this directly to performance: “No one can put in his best performance unless he feels secure.”

No one can put in his best performance unless he feels secure.— W. Edwards Deming

Security did not mean comfort or low standards. It meant people could tell the truth about obstacles, broken methods, confusing handoffs, and unreliable tools.

The nurse said the team did not need another campaign. They needed clearer refill rules, working equipment, and time to fix handoff problems. In other words, they needed management to improve the conditions of work.

Elena changed the plan. Northstar still measured access, callbacks, and patient experience, but the monthly meeting no longer ranked clinics. Managers studied variation, common barriers, and where the system made good work harder than it needed to be.

Each clinic selected one problem to study: a refill workflow, a daily equipment check, or message routing. The tone changed slowly. People began to speak more plainly about the system. The clinics improved unevenly, but honestly.

Northstar did not need to manufacture motivation with a contest. It needed to stop draining it.

Why we reach for pressure first

We reach for rankings and incentives because they seem concrete. They show seriousness and fit a common assumption: if people cared more, tried harder, or competed intensely, results would improve.

The difficulty is that much performance is shaped by the system. Tools, methods, patient mix, handoffs, training, and leadership habits all affect results. When we rank people without understanding those conditions, we may mistake system effects for personal merit.

Deming warned about the damage: “No one can enjoy his work if he will be ranked with others.”

No one can enjoy his work if he will be ranked with others.— W. Edwards Deming

Ranking changes the psychology of work. Instead of asking, “How do we improve this system?” people ask, “How do I avoid being at the bottom?” Instead of sharing learning, teams may protect their position.

We do this with good intentions. We want accountability and urgency. But when accountability becomes judgment without knowledge, it can weaken the very cooperation the organization needs.

Short-term thinking makes the habit more tempting. A contest or rating system can be launched quickly, while system improvement requires patience. We may mistake faster pressure for better leadership, especially when the effects on trust and learning are harder to see.

These habits do not only affect morale. They affect quality, cost, service, and trust. Preserving learning and pride in work builds a capability that is hard to copy.

What leaders can do instead

Deming’s point is not that leaders should ignore results. It is that they should understand how results are produced. Motivation improves when management removes barriers and helps people contribute to a clear aim.

* Begin with respect for purpose. Assume that most people want to do work they can respect. Start by asking what helps or blocks that desire before adding more pressure.

* Study the system before judging performance. Look for the conditions shaping results: methods, tools, handoffs, workload, training, and variation. A number is a starting point for learning, not a verdict.

* Remove fear from problem reporting. Make it normal for people to name obstacles without turning every conversation into an evaluation. Leaders need the truth more than they need polished explanations.

* Replace ranking with shared improvement. Use measures to understand the work, not to set people against one another. Cooperation improves when people can learn across boundaries without losing status.

* Protect pride in workmanship. Give people a real chance to do good work by clarifying purpose and removing recurring frustrations. Better service and steadier results grow from that capability.

Deming also warned: “The merit system destroys cooperation.”

The merit system destroys cooperation.— W. Edwards Deming

That is hard because merit systems feel fair on the s…

Zero defects sounds like seriousness. It sounds like standards. It sounds like the kind of phrase a responsible executive should say when quality slips.

That is exactly why it is dangerous.

The problem is not the desire for fewer defects. The problem is what happens when we turn that desire into a slogan, a target, or a public demand on people who do not control the system that produces the work. What feels like leadership can quietly become a substitute for leadership.

What the slogan hides from us

W. Edwards Deming’s criticism of zero defects is often misunderstood. He was not arguing for tolerance of poor quality. He was arguing against the managerial habit of demanding an outcome without changing the conditions that make the outcome possible.

That distinction matters in every industry. In manufacturing, it shows up in defect goals that do not address process capability. In software, it shows up in release pressure that ignores unstable requirements and weak handoffs. In safety, it shows up in signs that celebrate days since last injury while the underlying hazards remain in place.

We are drawn to slogans because they simplify reality. They give us something visible to say and something visible to measure. But the ease is deceptive. When the system stays the same, the number becomes the object of management, and the work of improvement gets pushed aside.

That is where the trouble starts.

What happened at Northstar Flow

Northstar Flow sold workflow software to mid-sized manufacturers. The company had hit a rough stretch. Three releases in a row had produced customer-facing bugs that should have been caught earlier. Support tickets were climbing. Sales was uneasy. The executive team wanted to show control, and fast.

At the Monday leadership meeting, the COO wrote four words on the whiteboard: Zero Defects Next Release.

The line had force. It was clean, memorable, and easy to repeat.

Within days, dashboards appeared. Teams were compared by escaped defects. Release reviews got tighter. People spoke more sharply. Product managers defended requirement changes. Engineers argued over classifications. Testers spent more time debating the count than learning from it.

Maya, who led product, felt the pressure immediately.

“We cannot do another release like the last one. Customers are tired of hearing that we are fixing it in the next patch.”

Daniel, the engineering leader, agreed with the urgency but not with the response.

“I agree. But the board on the wall is changing behavior. People are protecting the number.”

That was the turning point. The company had not become more capable. Requirements were still changing late. Test environments were still inconsistent. Handoffs between product, engineering, and support were still rushed. But now fear had entered the system in a more organized way.

At the next review, one team delayed logging a defect until after a release decision because no one wanted another mark against the group. Another team resisted a customer-reported issue by calling it a configuration problem until support escalated it twice. The visible count improved a little. The customer experience did not.

Deming warned directly against this kind of move: “Eliminate slogans, exhortations, and targets for the work force asking for zero defects and new levels of productivity.”

Eliminate slogans, exhortations, and targets for the work force asking for zero defects and new levels of productivity.— W. Edwards Deming

Once Maya and Daniel saw the pattern, the conversation changed. They stopped asking who had let the company down and started asking which conditions made escape likely. Late requirement changes were entering sprint work without a reliable review path. Regression coverage was uneven across older modules. Support was learning about release risk after key decisions had already been made.

They started with three changes. No release would be judged by a single defect number. Every release candidate would get a cross-functional review of requirement changes, test coverage risk, and support exposure. And escaped defects would be reviewed jointly, not to assign blame, but to separate recurring patterns from one-off events.

The next release was not perfect. But it was calmer. Fewer issues escaped. The ones that did appear were easier to trace. Support was prepared. Customers heard a clearer explanation. Trust began to recover because the company looked less frantic and more competent.

Maya said it plainly: “We finally look more serious now that we stopped promising perfection.”

And Daniel answered with the real shift in thinking: “Because now we are improving the work, not just demanding a result.”

Where managers get trapped

Most of us do not fall into the zero-defects trap because we do not care about quality. We fall into it because pressure makes visible promises feel like responsible action.

When numbers get worse, we want to show resolve. We want a message everyone can understand. We want the organization to know we are taking the problem seriously. So we set a target, publish a board, or attach consequences to the result. We tell ourselves that clarity will create performance.

Sometimes it creates compliance theater instead.

This is where Deming’s teaching is still unsettling. He forces us to admit that many of the outcomes we react to are produced by the system more than by individual effort. If we do not understand variation, we will treat every bug, delay, or accident as proof that someone needs more pressure. If we misunderstand incentives, we will reward the appearance of control while the underlying process stays weak.

Deming said it directly: “[Zero defect] exhortations only create adversarial relationships, as the bulk of the causes of low quality and low productivity belong to the system and thus lie beyond the power of the work force.”

[Zero defect] exhortations only create adversarial relationships, as the bulk of the causes of low quality and low productivity belong to the system and thus lie beyond the power of the work force.— W. Edwards Deming

That is not soft management. It is demanding management. It asks more from leaders, not less. It asks us to study the work, redesign the conditions, and remove the causes that keep reproducing the same disappointments. When we fail to do that, we do not just weaken internal performance. We weaken trust, reliability, and the kind of reputation that strengthens position over time.

What leaders can do instead

* Replace slogans with method. When performance slips, resist the urge to lead with a banner or a demand. Ask first what method will actually change the conditions producing the result.

* Study the system before judging the people. Look at requirements flow, handoffs, definitions, tools, timing, and feedback loops before concluding that effort or commitment is the problem.

* Separate recurring patterns from one-off events. A leader’s job is not to react emotionally to every data point. It is to learn whether the problem is built into the process or coming from a distinct special cause.

* Design reviews that improve learning, not fear. If a metric can be improved by hiding, reclassifying, or delaying bad news, the metric is teaching the wrong lesson. Build review routines that surface patterns early and safely.

* Define quality in customer terms. Conformance matters, but customers experience quality as reliability, clarity, fit, and trust over time. Improvement becomes more valuable when it strengthens those things, not just the internal count.

* Treat durable advantage as a consequence, not the aim. Better systems create steadier service, fewer surprises, and stronger confidence. Over time, that becomes hard for competitors to copy, but only because the management capability underneath it is real.

The leadership standard that matters

Deming's point was never that defects do not matter. His point was that demanding perfection is not the same as building capability. Even if the count improves for a while, leadership still has to improve the who…

Most leaders talk about employee retention as if it were mainly a hiring problem, a pay problem, or a culture problem. W. Edwards Deming points us somewhere more demanding. What if people leave because the system makes good work too hard, and honest work too risky?

If that is true, retention is not a side issue. It becomes a signal about whether management is preserving dignity, pride, and trust inside the work. And that signal matters long before a resignation lands on someone’s desk.

The real question behind who stays

In Deming’s view, people do not arrive at work empty. They come with curiosity, energy, and some desire to do a job well. Management does not create those qualities from nothing. More often, management either protects them or steadily crushes them.

That is why employee retention deserves deeper attention than it usually gets. When people withdraw, go quiet, or leave, we are often seeing the combined effects of system friction and damaged psychology. Conflicting priorities, weak handoffs, judgment-heavy reviews, and fear of speaking plainly can make even capable people feel trapped between doing the job and protecting themselves.

The usual leadership response is to ask how to make people stay. Deming would push us to ask a harder question first: what kind of management makes staying feel worthwhile?

That question becomes easier to see in a small company, where every resignation carries operational consequences. It also becomes easier to avoid, because leaders can tell themselves the issue is personal fit, labor market pressure, or attitude. A story helps make the distinction clearer.

What Lena finally saw in the resignations

Lena ran a growing service company with about thirty employees. Over the last year, three experienced people had left. Two newer hires were already interviewing elsewhere. Customers were beginning to notice uneven service, and Lena had settled on a simple explanation: people were becoming less committed.

So she responded the way many leaders do. She tightened expectations, increased pressure around the numbers, and added a pay increase with a retention bonus. For a week or two, the operation looked sharper.

Then the same problems returned.

Work was rushed. Mistakes repeated. One employee resigned with almost no warning.

Then Marcus, a team lead who rarely complained, asked for a private conversation.

“People aren’t leaving because they don’t care,” he said. “They’re leaving because it’s getting harder to do a good job and harder to say that out loud.”

Lena pushed back. She pointed to the changes she had already made.

“We made changes. We listened. I can’t just lower the standard because people feel pressure.”

Marcus did not argue about standards.

“This isn’t about lowering the standard,” he said. “It’s about what the work feels like now. Priorities change in the middle of the day. One manager says speed matters most. Another says not to miss a single detail. Suggestions disappear. And when the numbers look bad, people start protecting themselves.”

That conversation stayed with her because it explained more than turnover. It explained the silence. Questions were being delayed until problems became urgent. Small defects were being fixed quietly instead of discussed. People were cooperating less because the system had taught them that caution mattered more than candor.

Deming captured the psychological core of the issue in one direct line: “No one can put in his best performance unless he feels secure.”

No one can put in his best performance unless he feels secure.— W. Edwards Deming

Lena began to see resignations differently. They were not isolated decisions made by disconnected individuals. They were clues about the conditions people were working in.

At the next staff meeting, she stopped talking about commitment and said something else.

“If the work is getting in your way, I need to know. If our management methods are making it harder to serve customers well, that’s on us to fix.”

Marcus answered quickly. “Fix the handoffs first. That’s where the day starts going wrong.”

She did. Lena removed the quiet individual comparisons that had become rankings. She simplified priorities so people were not being pulled in opposite directions. She asked supervisors to surface recurring barriers and respond to them visibly instead of explaining them away.

The room did not become candid overnight. But people kept naming the same obstacles: missing information at handoff, last-minute changes, and reviews that felt more like judgment than help.

Deming named that danger clearly too: “Evaluation of performance nourishes fear.”

Evaluation of performance nourishes fear.— W. Edwards Deming

Once Lena could see the pattern, she stopped treating turnover like a mystery. She treated it like evidence. Within a few months, fewer people were talking about leaving. Problems reached supervisors earlier. Rework began to drop. Customers noticed steadier service because the work itself was becoming easier to do well.

And that mattered in the market. Not because Lena launched a retention initiative, but because better management was starting to produce more reliable service than nearby competitors could easily match.

Why we keep misreading turnover

Many of us were taught to read turnover at the level of the individual. We ask who lacked commitment, who wanted more money, or who was not resilient enough for the pace. Sometimes those factors are real. But when the pattern repeats, that lens becomes dangerously incomplete.

We miss the system that is shaping behavior.

We also underestimate how quickly fear changes the quality of information we receive. When people believe that bad news will be used against them, they soften it, delay it, or keep it to themselves. When performance reviews feel like judgment, people manage appearances. When priorities conflict, they choose self-protection over cooperation. From the outside, this can look like disengagement. Inside the system, it is often a rational response.

Deming’s point was that common reward and evaluation practices can drain intrinsic motivation and replace it with self-protection.

That is why superficial retention efforts so often disappoint. Bonuses, slogans, and urgent recruiting can help at the margin, but they do not remove the conditions that are pushing people away. If anything, they can deepen cynicism when employees are asked to care more while the system still makes good work unnecessarily hard.

When we react to resignations without studying the conditions behind them, we do not just weaken internal performance. We weaken learning, continuity, and long-term trust with customers. Over time, that becomes a competitive problem as well as a people problem.

What thoughtful leaders can do next

Deming’s aim was not merely to reduce fear. It was to create conditions in which people could contribute with interest, confidence, and pride. As he wrote: “[A leader] tries to create for everybody interest and challenge, and joy in work.”

[A leader] tries to create for everybody interest and challenge, and joy in work.— W. Edwards Deming

That is a demanding management standard. It means we cannot treat retention as a human resources metric detached from how the work is designed and led.

* Study the pattern, not the last resignation. Look at turnover alongside rework, absenteeism, customer complaints, overtime, and silence. Those patterns often reveal the recurring barriers that make people feel ineffective or unsafe.

* Remove fear where information should flow. Examine reviews, rankings, and judgment-heavy routines that teach people to protect themselves. Better information begins when people believe honesty will lead to improvement rather than punishment.

* Improve the work before asking for more commitment. Clarify priorities, repair handoffs, and respond visibly to recurring obstacles. People trust management more when they can see that leaders are trying to improve the process.

* Protect dignity as a management responsibility. Pride in workmanship is …

Most leaders would never say profit does not matter. The problem is almost the opposite. They talk about profit constantly. Budgets tighten. Targets multiply. Departments are pressed to improve their own numbers. On the surface, that can look like discipline.

But the deeper question is harder. If profit really matters, why do so many management habits reduce trust, increase waste, and make the organization less capable over time? That is the Deming challenge. Profit is real. It is necessary. But it is not managed well by chasing it directly.

Why chasing the number breaks the system

Deming’s view of profit is more demanding than the usual financial conversation. He did not treat profit as optional, but he did reject the idea that leaders can secure it by applying more pressure to visible figures. He saw profit as the result of better management of the whole system over time.

He put it bluntly: “Emphasis on short-term profit defeats constancy of purpose and long-term growth.”

Emphasis on short-term profit defeats constancy of purpose and long-term growth.— W. Edwards Deming

That sentence is uncomfortable because it names a pattern many organizations normalize. Under pressure, leaders narrow their time horizon. They defer maintenance. They cut learning. They treat quality work as a cost center. They ask each department to maximize its own result and assume the whole organization will somehow benefit.

It usually does not. And that is where the real trouble begins.

To see why, it helps to look at a story.

When every department wins and the organization loses

Riverview Health Network was under familiar pressure. Margins were tight. Labor costs were rising. Denied claims were getting more attention from the board. Senior leaders responded in a way many organizations would recognize: they asked each vice president to improve the financial performance of his or her own area.

Andrea, the chief operating officer, took the assignment seriously. She tightened staffing controls, pushed harder on throughput, and made departmental targets more visible. Radiology watched utilization. Registration watched speed. Billing watched collections. Clinic managers were told to monitor overtime closely.

When Marcus raised concerns early, Andrea answered the way many executives would.

“I understand that. But we cannot ignore the numbers. If every department improves its margin, the organization improves.”

For a short while, the reports looked better. Overtime dipped. A few local targets moved in the right direction. The monthly review felt calmer.

Then the strain showed up elsewhere. Patient complaints increased. Claims were denied because registration was incomplete. Nurses were calling managers about delays in imaging and discharge paperwork. Billing teams were spending more time on rework. Staff tension rose because every department was defending its own scorecard and pushing problems downstream.

Marcus, who led patient access, finally said what the system was already revealing.

“We are improving each piece on paper, but the whole thing feels harder to run.”

Later, standing at a whiteboard with the patient journey mapped from scheduling to billing, he made the problem even plainer.

“We are managing this like separate profit centers.”

That was the turning point. Andrea could see that no single department looked wildly broken on its own. Yet the system as a whole was producing delay, hidden cost, frustration, and lost trust.

At the next leadership meeting, she changed the conversation.

“We keep saying profit is the priority. But if that were really true, we would stop making decisions that increase total waste. We are protecting monthly appearances and creating bigger losses underneath them.”

The room went quiet. Then she took the next step.

“We need to manage patient flow, information quality, and cooperation across the system. We cannot ask each area to win separately and expect the whole network to win.”

Profit still mattered. But now she could see that the organization had been protecting appearances while creating bigger losses underneath them.

So Riverview stopped treating departmental targets as the main story. Leaders studied handoffs, duplicate work, and points where one team’s local savings created losses somewhere else. They reduced repeated data entry. They gave front-line teams time to improve coordination. They stopped rewarding savings that only looked good because another department absorbed the pain later.

Not every local measure improved immediately. Some looked worse before the whole system stabilized. But within a few months, denied claims fell, patient complaints eased, and financial performance became steadier because the organization was wasting less effort.

That is not soft thinking. It is better management.

Why we keep falling into this pattern

Most of us have worked inside systems that teach us to manage from the numbers backward. If the margin is down, squeeze harder. If costs rise, freeze spending. If one area looks weak, push that area to perform. We do not usually mean to damage the organization. We are trying to be responsible.

That is why this pattern is so persistent. It feels practical. It feels serious. It feels financially mature.

But when we react that way, we often confuse local measures with system performance. We treat symptoms as causes. We misread variation. We reward visible action even when it increases hidden waste. And because each team is pressed to defend its own result, internal competition begins to replace cooperation.

Deming saw the danger clearly: “A system must be managed. It will not manage itself. Left to themselves, components become selfish, competitive, independent profit centers, and thus destroy the system.”

A system must be managed. It will not manage itself. Left to themselves, components become selfish, competitive, independent profit centers, and thus destroy the system.— W. Edwards Deming

That does not just weaken internal performance. Over time, it weakens the organization’s position in the market as well. Customers experience the friction. Employees feel the strain. Rework consumes capacity. Trust erodes.

Meanwhile, an organization that manages flow, reliability, and cooperation is building something much harder to copy.

Four ways to care about profit more seriously

* Define the aim before chasing the numbers. Financial results matter, but they cannot be the only language of leadership. Clarify what the organization exists to do well for customers, patients, employees, and the future, then manage profit as a result of fulfilling that aim better.

* Read the system, not just the scorecard. When one number moves, resist the urge to react immediately. Ask what in the system is creating the result. Look at handoffs, rework, delays, and recurring failure points before you tighten pressure on any one group.

* Stop rewarding local wins that create total loss. A department can improve its own figures while making the whole organization slower, more expensive, and less trusted. Financial discipline becomes more real, not less, when leaders refuse savings that only shift cost somewhere else.

* Invest in capability while pressure is high. Training, redesign, better methods, and stronger cross-functional cooperation are often the first things leaders cut when margins tighten. Deming’s view is the reverse: those are the conditions from which healthier profit grows.

Deming wrote: “Profit comes from repeat customers—those that boast about the product or service.” Trust, reliability, and coordinated service do more than make an organization admirable. Over time, they strengthen its position in ways competitors struggle to match.

Profit comes from repeat customers—those that boast about the product or service.— W. Edwards Deming

What profit looks like in a better system

The management mistake is not caring too much about profit. The mistake is caring about it in a shallow way, as if harder pressure on visible numbers could substitute for improvement of the system that produces…

Leaders today rarely suffer from a lack of data. The deeper problem is that we often do not know what the data is asking us to do. A number rises, and we feel pressure to respond. A number falls, and we assume something worked. In both cases, we may be reacting to movement without understanding meaning.

Control charts matter because they help us separate ordinary variation from a real signal. That sounds technical. It is actually a practical discipline for calmer judgment, better decisions, and less wasteful management.

Why this changes the work of leadership

Control charts are often treated as a specialist’s tool, useful for analysts or quality teams but distant from executive work. W. Edwards Deming saw them differently. He treated them as a way for management to distinguish what belongs to the system from what points to something unusual.

That distinction changes the kind of leadership action that makes sense. If the chart shows a special cause, we investigate what changed. If the chart shows a stable but disappointing system, we stop chasing episodes and improve the design of the work itself.

Deming captured the idea in one memorable line: “The control chart is the process talking to us.”

The control chart is the process talking to us.— W. Edwards Deming

That is why the concept matters beyond reporting. A chart is not there to decorate a dashboard or make review meetings look disciplined. It is there to help us hear the system before we explain it, correct people for it, or reorganize around the latest fluctuation. A hospital story makes that distinction easier to see.

What St. Anne’s learned in one meeting

At St. Anne’s Hospital, emergency department boarding times had become a recurring source of executive concern. Week by week, the numbers moved up and down. Patients waited too long for beds upstairs, complaints kept coming, and senior leaders felt pressure to show that they were taking charge.

Elena, the chief operating officer, looked at the latest report and did what many capable leaders do under strain. She wanted urgency, accountability, and visible follow-through.

“I want each unit leader in here this afternoon. If a floor is holding patients too long, I want to know why. And I want targets by Friday.”

Marcus, the vice president of operations, had seen this pattern before. A bad week created urgency. A better week brought relief. Neither reaction was producing understanding.

Instead of bringing Elena another dashboard, he brought her a control chart. He had plotted six months of emergency department boarding times and discharge completion before noon. Elena studied the page for a moment and asked the obvious question.

“So what am I looking at?”

Marcus answered without technical jargon.

“Not just a trend line. This chart tells us whether we’re looking at the normal voice of the system or a signal that something unusual happened.”

That was the turning point. Most of the points were inside the control limits, with no unusual pattern. The process was stable, even though the performance was still not good enough. But two points clearly broke the pattern. Those were signals.

Elena leaned in. The weekly swings that had felt dramatic now looked different. Not like a fresh management failure every week, but like one repeating system interrupted twice.

“What caused the two signals?”

Marcus pointed to specific events. One week reflected a plumbing failure that reduced bed availability. The other reflected a cyberattack drill that slowed admissions and discharge orders. Those were special causes. They deserved investigation. But the larger boarding problem was built into the way the hospital was operating every day.

That is the managerial value of the chart. It did not excuse the delays. It clarified the level of action required.

Stable did not mean acceptable. It meant predictable under current conditions. Elena was no longer looking at a mystery that changed every week. She was looking at a system that was reliably producing an unsatisfactory result, with two real interruptions layered on top.

“So the chart is telling us two things at once,” she said. “Chase the signals. Improve the system.”

Exactly.

That afternoon’s meeting changed shape. Elena canceled the ranking discussion. Instead, she asked for a review of the two special-cause events and a separate cross-functional look at bed management, discharge timing, transport delays, and nursing handoffs. Over time, genuine disruptions were investigated faster, while chronic system problems became easier to name and improve.

That is how the problem began to resolve. The hospital stopped treating every fluctuation as a fresh crisis and started managing patient flow as a system.

Why we keep getting this wrong

Most of us do not misuse performance data because we are careless. We do it because pressure changes what feels responsible. When a number worsens, we want an explanation immediately. We want to know who owns the problem, what action will be taken, and how soon the result will move back in the right direction.

That instinct feels practical, but it often drives poor management. Much of the time, the result in front of us comes from the system’s ordinary behavior. Yet we treat a routine rise or drop as proof that something specific went wrong. Then, on other occasions, we miss a genuine signal because we have trained ourselves to regard every fluctuation as noise.

Deming was explicit about the stable case: “When a control chart indicates no special cause present, the process is said to be in statistical control, or stable. The average and limits of variation are predictable with a high degree of belief, over the immediate future. Quality and quantity are predictable. Costs are predictable.”

When a control chart indicates no special cause present, the process is said to be in statistical control, or stable. The average and limits of variation are predictable with a high degree of belief, over the immediate future. Quality and quantity are predictable. Costs are predictable.— W. Edwards Deming

That is the situation leaders face more often than they realize: a system performing exactly as it is currently designed to perform, even when the result is disappointing.

But Deming was equally clear about the less common case: “A point outside the control limits is a signal (an operational definition for action) of a special cause, which indicates the need for action—try to identify the special cause, and if it can recur, eliminate it.”

A point outside the control limits is a signal (an operational definition for action) of a special cause, which indicates the need for action—try to identify the special cause, and if it can recur, eliminate it.— W. Edwards Deming

Taken together, those two statements define the management problem. We get into trouble when we treat ordinary variation like a special event, or when we treat a real signal as just another routine fluctuation.

When we fail to see that, we tamper. We reshuffle priorities, pressure teams, explain every point, and make promises the system cannot yet keep. The cost is not just internal confusion. Over time, it also weakens service, trust, and the kind of dependable performance that becomes hard for competitors to copy.

What better leadership looks like

Before any leader turns a chart into a verdict on people or a call for hurried intervention, the first job is to understand what kind of variation the system is showing.

* Ask what kind of variation you are seeing. Before reacting, decide whether the result points to a special cause or to the normal behavior of the current system.

* Separate investigation from improvement. A signal calls for inquiry into what changed. A stable but unsatisfactory pattern calls for redesign of the work, not more pressure on individuals.

* Stop rewarding explanations without evidence. When we insist on a story for every movement in the numbers, we train managers to narrate noise instead of learning from the process.

* Treat predictability as a leadership asset. A stable p…

Many management teams are praised for speed. They launch new initiatives and talk about momentum as if motion itself were evidence of progress. But fast action without disciplined learning creates a different problem: we spread assumptions through the system before we know whether they are sound.

That is why W. Edwards Deming’s Plan-Do-Study-Act matters. It gives leaders a way to slow down certainty without slowing down improvement. In the long run, it produces better service, lower waste, and a steadier reputation.

Why leaders need more than a pilot

Plan-Do-Study-Act (PDSA) is often described as an improvement cycle. That is true, but it can sound smaller than Deming intended. PDSA is a way to connect theory, prediction, action, and learning.

Plan means more than choosing an idea. It means stating what you think is happening, what change you want to test, and what you predict will follow. Do means carrying out that test, usually on a limited scale. Study means comparing the result with the prediction and taking surprises seriously. Act means deciding whether to adopt the change, abandon it, or run another cycle with a better theory.

Deming put the underlying point simply: “Management in any form is prediction.”

Management in any form is prediction.— W. Edwards Deming

That is what many change efforts skip. We move from concern to action without ever being clear about the theory behind the action. Then we mistake activity for learning, or a short-term result for proof.

A story from commercial property management makes the problem easy to see.

What Harbor Point learned by slowing down

At Harbor Point Property Group, the executive team was under pressure. Tenants in three downtown office buildings were complaining about slow maintenance work, repeat visits, and weak communication from the service desk. Renewal season was approaching, and nobody wanted owners asking why routine service felt unreliable.

Claire, the head of operations, opened a Monday meeting with a familiar managerial move. She wanted speed, clarity, and a visible response.

“We need faster resolution times. I want every building manager under four hours for routine maintenance requests by next month.”

It sounded decisive. Complaints were rising. The pressure to look responsive was real.

But Jordan, the regional operations director, had spent the previous week reading work-order notes from the buildings. He saw something Claire’s demand did not explain. Some tickets stayed open too long. Others were closed quickly, then reopened. Vendor dispatches were inconsistent. Tenant descriptions were often incomplete. The pattern looked messy, not simple.

When Claire pressed him, Jordan answered with the line that changed the meeting.

“I think we know the symptom. I’m not sure we know the problem yet.”

That was the turning point. Instead of accepting a broad portfolio-wide push for faster close times, Jordan proposed a PDSA cycle. One building. One category of request. Two weeks. Plumbing calls in Franklin Tower only.

“Two weeks feels slow,” Claire said.

“Only if we confuse motion with learning,” Jordan replied.

This was the Plan stage, and he made it concrete. The service desk would ask three new intake questions before dispatching a plumber. Building staff would classify each request by severity. Vendors would receive tighter work orders with tenant access details and photos when available. Jordan’s prediction was clear: first-visit completion would improve, repeat visits would fall, and tenant updates would improve even if average close time did not improve right away.

That kind of planning is not paperwork.

It is disciplined thinking.

As Deming wrote: “Step 1 [Plan] is the foundation of the whole cycle.”

Step 1 [Plan] is the foundation of the whole cycle.— W. Edwards Deming

The Do stage followed. For two weeks, Franklin Tower used the revised intake method only for plumbing calls. The service desk logged the new questions. Building staff tagged urgency consistently. Jordan reviewed requests daily to make sure the test was being carried out as planned.

Then came Study. The headline result was mixed. Average close time improved only slightly. If Harbor Point had judged the test by a single visible metric, the effort might have been dismissed as disappointing.

But the rest of the evidence told a more useful story. First-visit completion improved sharply. Repeat visits fell. Complaints about poor communication dropped. And one surprise stood out: the biggest delays were not coming from the plumbers. They were coming from incomplete tenant access information and late approvals for after-hours entry.

Claire saw it immediately. The dispatch script had helped, but not in the way they first expected.

“Right,” Jordan said. “We learned more than whether the average moved. We learned where the friction actually is.”

That answer captured the real value of the cycle.

That led to Act. Harbor Point kept the stronger intake questions, added a clearer path for access approvals, and ran another cycle in a second building with a different tenant mix. The second round confirmed some of the original theory and corrected the rest. The intake method held up. The approval issue mattered even more than they first thought. One vendor adapted quickly; another needed coaching.

In time, Harbor Point did standardize parts of the process. But they did not do what the first meeting had nearly produced. They did not issue a broad demand to close tickets faster and hope for the best. They acted on what the cycles taught them. Tenant complaints fell. Repeat work declined. Service became more reliable.

That is what PDSA looks like in practice.

Not delay. Learning strong enough to justify action.

Where we usually go wrong

Most of us do not resist PDSA because we dislike learning. We resist it because pressure makes immediacy feel responsible. When complaints rise, costs increase, or customers get restless, we want to show movement.

That impulse is understandable. It is also risky.

We often confuse a visible response with a thoughtful one. We roll out a policy, tighten a target, or announce a new standard before we have stated the theory behind it. Then, when numbers move, we read the movement as proof. If the result looks better, we congratulate ourselves too early. If it looks worse, we abandon the effort too quickly.

In both cases, we may learn very little.

Another problem is that we study outcomes too narrowly. We look for one summary number to tell the whole story. But systems rarely teach in a single measure. A useful test may reveal that the real issue lies in a handoff, an approval path, a vendor interaction, or a classification rule.

This weakens internal performance. Organizations that learn poorly create rework, inconsistency, and distrust. Organizations that learn well build steadier service and stronger trust.

What leaders can do instead

* Make prediction explicit. In every serious improvement effort, ask what theory is being tested and what result is being predicted. If we cannot answer those two questions, we are not ready to learn from the effort.

* Start smaller than your urgency prefers. A limited test is not a retreat from action. It improves the quality of action by creating learning with less cost and disruption.

* Study more than the headline result. A test may miss its most visible target and still reveal something crucial about approvals, timing, handoffs, demand, or coordination. The lesson may sit beside the metric we first cared about.

* Treat surprises as valuable evidence. When results differ from the prediction, resist the urge to defend the original idea. The gap between expectation and reality is often where the system becomes visible.

* Use Act as a leadership decision, not a ritual ending. Adopt what clearly helped. Abandon what did not. Revise the theory where the test exposed weak assumptions, then run another cycle. That is how management capability grows.

The discipline behind better improvement

The misconception is easy to understand: if a problem is…

Many leaders think inspection is what protects quality. If defects slip through, the answer seems obvious: add another check, another review, another pair of eyes at the end. It feels careful. It feels responsible.

But that habit can quietly raise cost, normalize rework, and keep management from seeing the deeper problem. The real issue is not what we catch at the end. It is what our system keeps producing in the first place.

The management trap

One of the easiest mistakes in management is to confuse detection with improvement. When something goes wrong, we naturally look for a way to catch it sooner, sort it faster, or keep it from reaching the customer. That instinct is understandable. It is also incomplete.

A company can become very good at finding defects and still remain trapped in a weak process that keeps making them. W. Edwards Deming said it plainly: “[Using] inspection to improve quality is too late, ineffective, costly.”

[Using] inspection to improve quality is too late, ineffective, costly.— W. Edwards Deming

The force of that statement is easy to miss. He was not arguing against all inspection. He was arguing against the belief that inspection is where quality is achieved.

Quality is shaped upstream, in design, methods, training, maintenance, scheduling, and in the way management coordinates the whole system.

To see how easily leaders drift into the opposite habit, consider a small manufacturer that had become highly disciplined at catching defects and surprisingly tolerant of producing them.

A small manufacturer, a familiar pattern

Hartwell Fixtures made custom metal display racks for local retailers. It was a solid Main Street manufacturer with a good reputation and steady orders. Elena, the owner, took pride in the fact that every rack was inspected before shipment.

From a distance, that looked like discipline.

On the floor, it looked different.

Welds were sometimes rough. Powder coating occasionally bubbled. Mounting holes did not always line up. None of those issues alone threatened the business. But together, they created a constant drag on the work. Final inspection kept finding defects, and rework kept absorbing time, attention, and overtime.

When a shipment was late for the third time in a month, Elena walked into inspection and saw what had gradually become normal: carts full of rework, operators waiting for decisions, and inspectors arguing over borderline pieces.

“What’s the fastest way to get this back under control?” she asked.

Marcus, her operations manager, answered with the logic the company had been living inside for months.

“We are catching most of the bad units,” he said. “If we add one more inspector on second shift, we can clear the backlog.”

That answer was practical. It was also revealing.

More inspection had already been the answer for months. Yet the backlog remained. Scrap was up. Overtime was up. Customers were becoming less patient. Hartwell was not dealing with a few isolated mistakes. It was operating inside a predictable system.

Later that day, Elena and Marcus looked at the recurring defects together. One week the problem centered on drilling. Another week it was coating. Another week it was warped tubing from a supplier. The pattern moved around, but the burden stayed in the same place: at the end, where the company tried to sort, repair, and rescue what the system had already produced.

Deming captured that logic memorably: “Our system of make-and-inspect, if applied to making toast, would be expressed: ‘You burn, I’ll scrape.’”

Our system of make-and-inspect, if applied to making toast, would be expressed: ‘You burn, I’ll scrape.’— W. Edwards Deming

That was Hartwell’s system in miniature. Make the rack. Find the defect. Grind it. Redrill it. Recoat it. Expedite it. Apologize for it. At some point, the company had confused recovery with quality.

That realization changed the conversation.

“If inspection is our main defense,” Elena said, “then we are planning to make defects.”

“Then where do we start,” Marcus asked, “if not at the end?”

Instead of asking how to strengthen the inspection wall, Elena and Marcus started tracing the defects upstream. They found fixture wear at the drilling station. They reviewed variation in incoming tubing from one supplier. They discovered that a setup shortcut had become normal on busy days. They also saw coating problems rise when rushed scheduling changes caused parts to sit too long between steps.

Inspection did not disappear. But it changed purpose. It became feedback about the process, not the company’s main theory of quality.

Marcus began tracking defect patterns to learn where the system was unstable. Supervisors stopped treating rework totals as proof that quality control was working. Elena stopped celebrating heroic saves that depended on overtime and last-minute sorting.

The result was not perfection overnight. Some defects still appeared. But rework began to shrink. Lead times became more predictable.

Inspectors spent less time debating borderline pieces. Operators had clearer standards and better equipment. Supplier conversations improved. The same people who had been blamed for defects started taking pride in racks that moved through without repair.

Hartwell did not improve because it got better at catching defects. It improved because management stopped pretending that catching defects was the same thing as creating quality.

Why we fall into this

This pattern is common because it flatters our instincts.

When a defect shows up, we want an immediate answer. We want action we can see. An extra inspection point, another signoff, a tighter approval step, or a fresh reminder to be careful all feel like responsible leadership.

They create motion. They create reassurance. They also let us avoid the harder work of asking what in the system keeps generating the same trouble.

We struggle here because symptoms are visible and systems are not. Rework has a location. Scrap has a number. Inspection has a department.

But the causes are often spread across design, training, maintenance, scheduling, supplier relationships, and unclear methods. No single problem screams for ownership, so we manage what we can see.

We also struggle because detection feels safer than redesign. Catching a defect at the end seems concrete. Improving the process that made it requires thought, patience, and cross-functional cooperation. It asks more of management.

That is why Deming’s reminder matters so much: “The quality of the product is the responsibility of management, working with the customer.”

The quality of the product is the responsibility of management, working with the customer.— W. Edwards Deming

If we hand that responsibility downward to operators or inspectors alone, we divide accountability in exactly the place it must remain integrated.

The difficulty is not that leaders do not care. It is that we can easily mistake visible control for actual improvement.

What leaders can do instead

* Redefine what inspection is for. Inspection can provide useful feedback, but it should not be your main strategy for achieving quality. Treat it as a way to learn about the process, not as proof that the process itself is sound.

* Follow defects upstream. When the same kinds of problems keep appearing, resist the urge to respond only at the point of discovery. Ask what in design, methods, materials, training, maintenance, or scheduling is making those outcomes likely.

* Stop rewarding recovery more than prevention. Heroic saves feel admirable, but they can hide an unhealthy system. Leaders should be careful not to praise overtime, sorting, and rework more than the quieter work of building stable flow and capable processes.

* Keep accountability where it belongs. Operators and inspectors can contribute insight, but they do not control the whole system. Management does. That means leaders have to coordinate across functions instead of treating quality as a department or a final checkpoint.

* Remember the business consequence. A system that produces dependable …

In many organizations, the phrase “use common sense” sounds perfectly reasonable. A mistake happens, a customer complains, or a process fails, and the instinctive response is to remind people to slow down and think.

But this familiar management reflex can quietly prevent improvement. When leaders rely on “common sense” explanations, they often focus on the individual closest to the problem instead of the system that produced it.

W. Edwards Deming warned that this habit does more than miss the cause—it can keep organizations trapped in the very patterns they are trying to fix.

Why “common sense” fails in management

Most managers have experienced the moment when something goes wrong. A customer receives the wrong order, an appointment is missed, or a deadline slips by.

The explanation appears obvious: someone made a mistake. Our instinct is to correct the person involved—remind them to be careful, encourage better judgment, or send a note to the team about paying closer attention.

These responses feel practical because work is done by people. But Deming argued that most recurring problems do not originate with individual effort or attention.

They are produced by the way work is designed—the methods, priorities, handoffs, and pressures that shape everyday decisions. When leaders overlook that reality, the same cycle repeats: correct the person, see temporary improvement, and then watch the problem return.

A small service company illustrates how easily this pattern develops—and what changes when a leader begins looking at the system instead.

A scheduling problem that kept returning

Maria owns a home services company that schedules technicians for repairs and installations across her city.

Over several months, customer complaints began to increase. Appointments were occasionally missed, technicians sometimes arrived without the right parts, and a few customers reported waiting all day for a visit that never appeared on the schedule.

One afternoon a customer called after waiting five hours for a technician who never arrived. Maria reviewed the call recording and quickly discovered the problem: the job had been placed into the wrong time slot.

It looked like a simple scheduling error.

Later that day she spoke with her operations supervisor, David.

“This one should have been obvious,” Maria said. “People just need to slow down and use some common sense when they’re entering these jobs.”

David agreed the mistake appeared straightforward, and the team reminded dispatchers to double-check their entries. For a short time the complaints seemed to ease.

But two weeks later another scheduling problem surfaced. Then another.

While reviewing scheduling logs, David noticed something unusual. The same type of error appeared across different dispatchers and across different shifts. It did not look like one employee being careless.

The team began examining the scheduling process itself. Service requests arrived through phone calls, website forms, and callbacks from technicians in the field.

The information customers provided varied widely, and dispatchers often had to guess which technician should handle a job. At the same time they were expected to answer calls quickly while entering appointments into the system.

During busy periods dispatchers were juggling two demands at once: respond to customers immediately and figure out incomplete job details. The errors appeared most often when call volume spiked and dispatchers rushed to keep up.

Deming described this common management reaction in The New Economics: “Common sense [mistakenly] tells us to speak to the operator about it when a customer reports something wrong with a product or with a service. ‘We have spoken to the operator about it; it won’t happen again.’”

Common sense [mistakenly] tells us to speak to the operator about it when a customer reports something wrong with a product or with a service. ‘We have spoken to the operator about it; it won’t happen again.’— W. Edwards Deming

Maria realized her earlier response had followed exactly that pattern. She corrected the person closest to the problem while leaving the process unchanged.

The team redesigned the scheduling system. They standardized intake questions so dispatchers received consistent information, clarified which technician handled each type of job, and adjusted call targets so dispatchers were not forced to rush scheduling decisions.

Within weeks the number of scheduling problems began to fall—not because employees suddenly became more attentive, but because the system guiding their work had improved.

As Deming wrote: “Action taken today may only produce more mistakes tomorrow. It may be important to work on the process that produced the fault, not on him that delivered it.”

Action taken today may only produce more mistakes tomorrow. It may be important to work on the process that produced the fault, not on him that delivered it.— W. Edwards Deming

Why leaders blame people first

Most of us do not intend to blame people unfairly. When something goes wrong, we simply want the problem fixed quickly.

But this instinct often leads us toward the most visible explanation instead of the most accurate one. When a failure appears, the person closest to the event becomes the natural focus of attention.

We see the dispatcher entering the wrong time slot, the technician forgetting a part, or the salesperson skipping a step. Because the action is visible, it feels like the cause.

What we often overlook is the system surrounding that moment—the incomplete information, conflicting priorities, or time pressure that shaped the decision.

Deming repeatedly warned leaders about relying on intuition alone. As he wrote in Out of the Crisis: “Best efforts are essential. Unfortunately, best efforts, people charging this way and that way without guidance of principles, can do a lot of damage.”

Best efforts are essential. Unfortunately, best efforts, people charging this way and that way without guidance of principles, can do a lot of damage.— W. Edwards Deming

Without a method for studying the system, we naturally react to effort instead of design. We ask people to try harder, be more careful, or use better judgment.

Sometimes that produces a short burst of improvement. But if the system creating the conditions remains unchanged, the same patterns quietly return.

What leaders can do instead

* Study the system before reacting. When a problem occurs, pause before correcting the individual involved. Examine the conditions that made the outcome possible and look for gaps in the process.

* Look for patterns instead of isolated mistakes. A single incident rarely explains much. Recurring problems across time, shifts, or teams often reveal systemic causes that individual events cannot explain.

* Reduce the need for judgment calls. Many service errors occur when employees must interpret incomplete information or conflicting priorities. Clear methods and operational definitions reduce variation.

* Align incentives with thoughtful work. When speed targets or short-term metrics conflict with careful decisions, people naturally respond to the pressure they feel most strongly.

* Improve the design of work continuously. Sustainable improvement comes from refining the system itself—clarifying processes, simplifying decisions, and removing barriers that prevent people from doing good work.

The system behind the mistake

Organizations depend on the judgment and effort of the people doing the work. But judgment works best when the system around it provides clarity, stability, and guidance.

When leaders move beyond “common sense” explanations and begin studying the systems that shape everyday decisions, improvement becomes far more predictable.

Instead of repeatedly correcting individuals, leaders begin redesigning the conditions that produce better outcomes in the first place.

That shift—from reacting to problems to improving systems—lies at the heart of Deming’s management philosophy.

A bad system will beat a good person every time.— W. Edwards Deming

…

Most organizations rely on annual performance reviews to evaluate contribution, allocate rewards, and create accountability. The logic feels straightforward: measure results, rate people, and recognize the strongest performers. For decades, this ritual has been treated as a basic tool of management.

But what if the very practice meant to improve performance quietly prevents real improvement from happening?

W. Edwards Deming believed annual performance reviews were not merely ineffective. He argued they were one of the most damaging management practices in modern organizations because they direct leadership attention toward judging individuals instead of improving the system that produces results.

To understand why, we have to rethink what actually creates performance in the first place.

Why Deming challenged performance appraisals

Leaders want to understand how well their organizations are performing. That instinct is healthy; good leadership requires visibility into results and a clear understanding of where improvement is needed.

Annual performance reviews promise a structured way to do this. They compress a year of work into scores, ratings, and rankings that guide compensation, promotion, and recognition.

But Deming argued that this approach misunderstands how organizations actually produce results. He wrote: “Basically, what is wrong is that the performance appraisal or merit rating focuses on the end product, at the end of the stream, not on leadership to help people.”

Basically, what is wrong is that the performance appraisal or merit rating focuses on the end product, at the end of the stream, not on leadership to help people.— W. Edwards Deming

In other words, reviews judge outcomes after the work is finished rather than improving the conditions that produce those outcomes in the first place. This difference—between judging results and improving the system that creates them—sits at the heart of Deming’s philosophy of management.

To see how this dynamic unfolds in practice, consider the experience of a school district wrestling with teacher evaluations.

A school district confronts the problem

In the Brookfield School District, evaluation season arrived every spring with predictable tension.

Teachers prepared documentation of their work while principals conducted classroom observations. District administrators compared performance scores across schools, and those numbers shaped pay increases, promotions, and professional reputations.

Marcus Lee, principal of Brookfield Middle School, had participated in the process for years, and each cycle followed the same pattern. Teachers worried about their scores, principals debated ratings, and district leaders reviewed charts comparing one school to another.

Yet the classrooms themselves seemed to change very little.

During a district leadership meeting, Marcus raised the concern with Superintendent Elena Ramirez.

“We keep having the same conversations,” he explained. “We review the ratings, we talk about who did well and who didn’t. But the classrooms themselves aren’t improving much.”

Ramirez understood the frustration, but she also saw the system as necessary.

“The reviews help us identify our strongest teachers,” she said. “Without them, how do we know who is performing well?”

Marcus paused before answering.

“That’s the problem,” he replied. “We think the scores explain performance. But most of the time they reflect the conditions teachers are working in.”

He pointed to several examples. Some teachers had consistent collaboration time with colleagues, while others rarely had time to work together. Some classrooms included far more complex student needs, and others had significantly more curriculum support.

The more Marcus studied the situation, the more he saw a pattern emerging.

As evaluation season approached, teachers became cautious. Collaboration slowed, and fewer people experimented with new lesson ideas because trying something new carried personal risk when results were judged individually.

The system was doing exactly what it was designed to do: judge individuals.

But something else was happening as well. Teachers began protecting their own standing rather than sharing openly, and leaders spent hours debating scores instead of studying the conditions shaping learning—curriculum support, scheduling, classroom composition, and collaboration time.

Slowly, the conversation shifted away from improving teaching and toward explaining ratings.

Deming warned about this dynamic decades ago: “Merit rating rewards people that do well in the system. It does not reward attempts to improve the system.”

Merit rating rewards people that do well in the system.It does not reward attempts to improve the system.— W. Edwards Deming

That comment stayed with Ramirez after the meeting. If the ratings were not revealing true performance, what should leadership be studying instead?

The answer emerged as district leaders began examining the system around teachers rather than the teachers themselves. They studied scheduling practices, collaboration time, curriculum support, and class composition. As these conditions became visible, many apparent differences in “performance” began to make sense.

The district gradually shifted its approach. Instead of relying on a single high‑stakes annual score, principals began holding ongoing coaching conversations with teachers, and teams studied classroom practices together while sharing what they were learning.

Over time the tone inside the schools began to change. Less energy went into defending ratings, and more attention went toward improving how teaching actually happened.

Why leaders keep returning to ratings

Most leaders adopt evaluation systems for understandable reasons. We want fairness, accountability, and a clear way to recognize strong contribution. Annual reviews appear to offer all three.

But Deming believed the deeper issue lies in how we interpret performance itself.

We often assume differences in results are primarily caused by individual effort or ability. When outcomes vary, we instinctively look for the person responsible, and the evaluation system becomes the mechanism for judging those differences.

In reality, most variation in performance within organizations is produced by the system people work within—its processes, resources, training, incentives, and leadership practices. When leadership focuses primarily on judging individuals, these systemic influences remain largely invisible.

The result is a familiar management cycle. Leaders react to outcomes rather than studying causes, and teams debate individual ratings instead of examining the conditions producing those results.

Over time, the evaluation process begins shaping behavior. People manage appearances, protect their own standing, and hesitate to take risks that might affect their ratings.

From a systems perspective, these responses are entirely predictable because the structure of the evaluation system quietly teaches people how to behave.

What systems-oriented leadership looks like

If judging individuals rarely produces improvement, where should leaders focus instead?

Deming’s answer was clear: study the system.

Leaders who want better performance begin by understanding the conditions shaping the work.

1. Study the system before judging the individual. When results differ, examine the environment in which people are working. Work design, training, tools, workload, and incentives often explain far more variation than most organizations realize.

2. Move leadership attention upstream. Evaluation systems tend to examine results after the work is finished. Systems-oriented leadership focuses earlier in the process—how work is designed, how knowledge flows, and where obstacles appear.

3. Create environments that support learning. Improvement requires experimentation and honest discussion of problems. Organizations improve faster when people feel safe sharing what they are discovering and where work is …