The org chart describes a job. Plan, organize, coordinate, control. Set the direction, then steer the organization toward it. It is a calm job. It is a reflective one. The manager sits a little above the work, sees the whole of it, and allocates.

Almost nobody does that job.

Henry Mintzberg established this in 1973, and the finding has never been seriously refuted. He did not theorize about managerial work. He watched it. He followed managers through their actual working days, recorded what they did minute by minute, and compared it to what the textbooks said they were supposed to be doing. The two had almost nothing in common.

That gap is the subject of this piece. It is also, for a great many capable operators, the exact place a career goes quiet. Not because they stopped being good. Because they were promoted into a job the textbook describes and the building does not contain.

This is a re-read of The Manager’s Job: Folklore and Fact — Mintzberg’s 1975 distillation of his observational study, reprinted by the Harvard Business Review in 1990 and largely unread since. The folklore he named is still taught. The facts he found are still true. The distance between them still ends careers.

What Mintzberg actually watched.

Read more

Most senior operators in regulated industries have been through an audit that came back clean and a failure that came shortly after. The two events do not appear to belong to the same operation. The audit looked at the documented system and signed it off. The failure exposed the operating system and named what was actually missing. The post-mortem walked back through both documents and made the gap legible. By then the damage was done.

This is the most common shape of a serious regulatory event. It is also the most underexamined. The standard reading is that the audit was inadequate, or the auditor missed something, or the operation got unlucky. None of those readings is structurally honest. The audit did exactly what it was designed to do. It tested the documented management system. The documented management system was not the operation. The two had drifted apart over years, and nothing in the audit was designed to detect the drift.

The recurring observation, across three weeks now, is structural. The artifacts are not the system. The slogans are not the work. The binder is not the operation. Each of these is a version of the same gap, named at a different scale. This piece names that gap at a third scale, the management system itself and asks the question most management systems are never asked.

Is your management system engineered to pass audits, or to detect its own drift?

The two are not the same. They feel like the same project until they aren’t. The difference shows up in the post-mortem and almost nowhere before it.

Two documents. One operation.

Every audited organization has two documents. They are rarely written down as two. They are usually treated as one. Naming them as two is the first move.

The first document is the binder. The procedures, work instructions, evidence files, training records, sign-offs, calibration logs, change-control entries, deviation reports, CAPA records, management-review minutes. The binder is what the auditor reads. It is also what most senior leaders believe the management system is.

The second document is the operation. The actual flow of work as performed by the people performing it. The shortcuts that emerged when the equipment changed. The tacit knowledge the long-tenured operator carries about which step actually matters. The workaround installed three years ago when the upstream system changed and the procedure didn’t. The way the night shift handles the variance the day shift doesn’t see. The operation is what the binder is supposed to describe.

In a healthy organization the binder and the operation are the same document. The binder describes the operation; the operation matches the binder; the two are refreshed together as conditions change. The auditor and the operator read the same thing, because there is only one thing.

This is rare. Most organizations have two documents that started out the same and slowly diverged. The divergence is not dishonesty. It is not the operators cutting corners or the managers hiding gaps. It is the natural physics of any documented system in operation. The world changes. The work changes. The document changes more slowly, because changing the document costs something, a sign-off cycle, a training update, a quality review. The work does not wait for the document. So the document falls behind. So the operation runs on knowledge the binder does not contain.

Once the two documents have diverged, the management system has split. The auditor’s system is the binder. The operator’s system is whatever the floor is actually doing. Both can be coherent in isolation. Both can pass internal review. The two of them together cannot survive a real test.

The post-mortem after a serious event almost always reveals the split. The procedures everyone signed off on were not the procedures the operation was running on. Nobody noticed because no audit was looking for the gap. The binder was being audited; the operation was being audited; the gap between them was not.

How the drift compounds.

The drift is not random. It moves in predictable directions, and it accelerates over predictable cycles.

The audit cycle itself is the largest mechanism. Most audits are scheduled. The schedule is known. The areas the auditor will examine are largely known. In the weeks before the audit, the binder gets refreshed in the places the audit will look. The training records get updated. The deviation reports get closed. The management-review minutes get drafted. The places the audit will not look stop getting refreshed.

This is not corruption. It is the rational response of any organization to a known examination. Effort flows toward the test. Effort flows away from everything else. Over enough cycles, the binder becomes a record of what was prepared for audits and very little else. The operation continues, somewhere underneath, running on a separate document that no audit cycle has touched in years.

The second mechanism is the operator’s calibration. Long-tenured operators learn, through experience, which procedures actually matter and which exist on paper because some past consultant put them there. They follow the ones that matter. They navigate around the ones that don’t. When an auditor arrives, they perform the procedure that’s documented. When the auditor leaves, they return to the procedure that works. They are not lying. They are running two systems and choosing the right one for the moment.

The senior leader who has never been an operator often misreads this. They see the audit-clean performance and conclude the system is working. The operator sees the audit-clean performance and concludes the senior leader does not yet understand which procedures are which. Both are operating in good faith. Both are looking at different documents.

The third mechanism is change without redocumentation. Every operation is in continuous low-grade change. Equipment gets replaced. Suppliers shift. Software updates. Personnel turn over. Each change touches one part of the operation. Almost none of the changes generate a corresponding update to the binder, because the binder is treated as a stable artifact rather than a live document. The operation is twelve percent different than the binder describes; then fifteen percent; then twenty. The auditor cannot see the gap because the auditor was reading the binder. The senior leader cannot see the gap because the senior leader is reading the metrics. The gap accumulates in the space neither of them is examining.

The fourth mechanism is the most dangerous. Once the drift has compounded for long enough, the binder and the operation become two cultures. The compliance function defends the binder. The operating function defends the work. They have separate meetings, separate language, separate priorities. When a real failure surfaces, the compliance function points at the procedure and asks why the operator did not follow it; the operating function points at the work and asks how anyone could have followed it. Both are right. The split was structural. Nobody had a mechanism to repair it because nobody had a mechanism to detect it.

By the time a regulator arrives in response to a real event, the gap is not a finding. The gap is the system the organization has been running for years. The finding is what the regulator names, but the finding is downstream of the architecture. The architecture chose this outcome long before the regulator wrote it down.

Four mechanisms that detect drift.

Read more

This is not a credibility problem. It is a legibility problem. Drucker was so widely quoted that most operators stopped reading him. The slogans replaced the substrate. And the slogans are not wrong, exactly. They are degraded. They kept the shape of the idea and lost the load-bearing structure underneath it.

The result is that an entire generation of executives learned a version of Drucker that sounds right and performs well in a strategy deck but cannot hold weight in an operating room. The version that holds weight is still in print. It is on the shelf behind you. The re-read is the work.

This piece walks three Drucker ideas that the corporate world got wrong. Not because Drucker was unclear.

Effectiveness is not efficiency.

Drucker drew a sharp line between effectiveness and efficiency. The corporate world erased it.

In The Effective Executive, Drucker defines effectiveness as the discipline of getting the right things done. Not the most things. Not the fastest things. The right things. He is specific about why this matters: the executive’s contribution is not measured by effort, by hours, by volume of output, or by the elegance of the process. It is measured by whether the right work happened and produced a result the organization needed.

Efficiency, by contrast, is doing things right. It is the process question. It answers: given that we are doing this work, are we doing it with the least waste? That is a useful question. It is not the executive’s question. The executive’s question is upstream: should we be doing this work at all.

The corporate world collapsed this distinction sometime in the early 1990s. “Effectiveness” became a synonym for “productivity.” The two words began appearing interchangeably in consulting decks, performance reviews, and management training. Once they merged, the executive’s real job disappeared. The real job is deciding what work matters. The replacement job is optimizing whatever work is already happening.

You can see the merger in the language. “Effective teams” now means “productive teams.” “Effective leadership” now means “leadership that delivers results.” Neither usage carries Drucker’s original weight. Drucker’s effective executive is not the one who produces the most. The effective executive is the one who could sit in a room, look at the nine initiatives underway, and name the two that actually matter. Then kill the other seven. That act of choice, not of optimization, is what Drucker called effectiveness.

Read more

The board went up. The colored cards moved for a quarter. The operations director left, or the program was rolled into the next initiative, or the manufacturing manager retired. Within eighteen months the board was either decoration or gone.

The consultants who installed the tools did not fail. The tools were never the thing that needed to be installed.

What Toyota built is not what most copiers think they were copying. The mistake is older than your career, and it explains the failure pattern you have been watching for thirty years. The pattern is so consistent that it stopped being surprising sometime in the 1990s, and yet most organizations are still running the same broken import on whatever the latest version is. Lean. Six Sigma. Operational excellence. Toyota Kata. Each new wave brought a new artifact and a new vocabulary. The substrate stayed missing.

This piece names the substrate. It does not teach the tools. It does not celebrate Toyota. It diagnoses what got copied, what got abandoned, and how to tell, in your own operation, by Tuesday which side of the line your improvement program sits on.

TPS is a management system. It is not a toolkit.

The famous tools are artifacts. Kanban, andon, jidoka, heijunka, 5S, A3, gemba walks. They look like the system. They are not the system.

The system is the integration. Standards become evidence. Evidence becomes escalation. Escalation becomes correction. Correction becomes new standard. The loop runs without interruption, every shift, every day, for decades. Every artifact you can name on a Toyota floor is a node in that loop. None of them functions outside it.

Take kanban. A kanban card is a signal. It tells one station that the next station has consumed inventory and is ready for more. By itself it is a piece of cardboard. What makes it work is the rest of the system. Heijunka (level production) determines what the upstream stations are allowed to build. Jidoka (build-quality-in) ensures that what they build is not defective. Andon (alert) makes any deviation visible within seconds. The standards behind each station make deviation legible at all. The escalation discipline guarantees that a pulled andon cord reaches someone who can actually decide. The hansei practice guarantees that the decision becomes a new standard, not just a fix.

Pull a kanban card out of that system. Put it on the wall of an operation that has none of the other elements. You have not implemented anything. You have decorated.

This is the management-system insight Deming spent forty years trying to make legible. A system is what makes individual practices coherent. Tools are the visible expression of the system; they are not its substance. Statistical process control is not a chart on a wall. It is the integration of measurement, standard, decision-right, and response. Strip out the integration and you are left with a chart, and a chart is not a control. Deming called it the deadly diseases of management. He named seven. Most American factories in the 1980s and 1990s caught at least four.

Most American factories in those decades tried to import Toyota’s tools without importing Toyota’s system. They named the result Lean. They watched it not compound. They blamed implementation. The implementation was not the problem. The architecture was.

The error is structural, not motivational. Plenty of the people running those programs were intelligent and committed. They were trying to solve a problem the program was incapable of solving, because the program was an artifact set, not a system. They did the best work the import allowed. The import did not allow the work that mattered.

In governance terms, this is the difference between passing an audit and being in control. A binder of procedures can pass an audit. Being in control means standards are alive, evidence is current, escalation is reflexive, and correction is faster than drift.

Most regulated industries have lived this directly. The audit gets passed. Then a real failure occurs, and the post-mortem reveals that the procedures everyone followed were never the procedures the operation was actually running on. The procedures were the artifact. The operation was the system. They had drifted apart years earlier and nobody had a mechanism to notice.

Toyota built the mechanism. Almost no one copied the mechanism.

The copies imported artifacts and abandoned substrate.

Here is what got copied, in roughly chronological order. Kanban, then 5S, then andon, then A3, then gemba walks, then “The Toyota Way” as a fourteen-principle framework, then “kata” as a habit-formation method. Each wave of Lean adoption picked a new artifact and called it the missing piece. Each wave dissolved on roughly the same timeline.

Here is what did not get copied, in the same chronological order. The long-term employment contract that let Toyota invest in a worker’s fifty-year career. The supplier relationships measured in decades, where the supplier and the customer shared engineers and engineering data and reciprocal obligation. The practice of “genchi genbutsu”, the senior leader going to the floor every day, often unannounced, often without the briefing. “Hansei,” the institutionalized practice of reflection on failure before celebration of success. The respect-for-people pillar that sits next to continuous-improvement in TPS’s actual stated architecture, not as a slogan but as a structural constraint on what the system is allowed to ask of the worker. The underlying willingness to improve slowly for sixty years before declaring anything finished.

The copies got the form. They missed the function. They got the chart on the wall. They missed what the chart was doing.

There is a structural reason for this, and it is worth naming honestly. Ackoff had it right. The question Western executives asked was: “how do we get Toyota’s results faster.” That question is structurally impossible to answer. Toyota’s results came from slowness. Toyota’s results came from compounding small adjustments over thirty thousand shifts. The right question was never how to accelerate the program. The right question was why the results required patience the Western firm did not yet have. Almost no one asked the right question. The wrong question was easier to staff against, easier to budget for, easier to put on a quarterly review.

Most “Lean transformations” were not transformations. They were attempts to extract speed from a system that paid Toyota in patience. The extraction failed because the system does not yield speed when you strip out its substrate. It yields what looks like Toyota for one quarter and then dissolves.

There is a quieter point here about respect for people, which is the principle Western adopters cut first. Kaizen — continuous improvement — sits in TPS alongside respect for people. The pairing is structural. Improvement without respect becomes coercive efficiency extraction. Respect without improvement becomes drift. Toyota built the place where both pillars held. Most copies kept the improvement language and silently dropped the respect language, then wondered why kaizen sessions started feeling like cost-cutting exercises with performance-review consequences.

The kaizen sessions felt that way because that is what they became. The substrate was missing.

When you read the failed implementations carefully, the pattern is the same. GM in the 1980s. The early NUMMI struggles. The Lean Six Sigma corporate programs of the mid-2000s, where green-belt certifications proliferated and operational results did not. The tools went up. The substrate did not exist. The artifacts decorated the operation for a quarter or three, then dissolved as soon as the operations VP changed.

NUMMI is worth pausing on, because it is the case most often cited as proof that TPS can work in a Western context. The full story is more interesting than the citation. NUMMI worked when GM and Toyota co-managed it because both organizations contributed substrate. GM provided the union relationships and the local supplier base. Toyota provided the system and the patience. When the partnership ended and Toyota walked away, NUMMI did not survive on the GM side. The system was real while two parents fed it. It dissolved when only one parent remained, because the GM parent had never absorbed the substrate; it had only co-existed with it.

This is the test no failed copier ever ran on themselves. “Do we have the substrate to feed this system after the consultants leave.” In almost every case the honest answer was no. Almost nobody asked the honest question.

The Lean industry survived this pattern by changing vocabulary every six or seven years. When the language got tired, the consulting firms rebranded. The artifacts stayed roughly the same. The substrate stayed roughly missing. The next executive cohort, fresh to senior responsibility, signed off on the next implementation, and the loop ran again.

Three readable signs of an artifact-only implementation.

If you have lived inside a Lean program, you already know what an artifact-only implementation looks like. You probably could not name why it felt off. The diagnosis lives in three signs. You can run them against your own current operation by Tuesday morning.

One. The visible artifacts exist, but the standards behind them are out of date.

Kanban cards in slots that do not match the latest takt time. Andon cords no operator has pulled in six months. 5S audits scored on whether the floor looks tidy, not on whether the labels match how work actually flows now. Standard operating procedures last revised when the line manager who wrote them was still employed. If the artifacts are present but the standards under them are stale, you do not have a system. You have decoration with a reporting layer. The reporting layer hides the staleness because the report formats were also frozen when the standards were.

The test: pull a random standard from any visible artifact. Compare it to the way the work is actually done today, with the operator who actually does it. If the answer is “we do it differently, but the standard hasn’t been updated,” you are not running TPS. You are running theater with TPS props.

Two. The improvement events end at the event.

Kaizen weeks happen. A3s get filled in. Reports get circulated. Posters get printed. Then nothing changes about how the next month’s standards are set. The improvement was the project; the project was the improvement. In a real TPS-style system, the kaizen week is a node. The standard it produces enters the system. The next loop changes shape. In a copied system, the kaizen week is a deliverable. Nothing flows from it because nothing was built to receive the flow.

The test: pick a kaizen event from twelve to eighteen months ago. Find the standard it produced. Find every shift since then where that standard was enforced, measured, and updated. If the standard exists on paper but does not appear in any active measurement loop, the event was decoration. The system is not running.

Three. The senior leaders do not go to the floor.

They schedule gemba walks as quarterly events with stage-managed routes. They are briefed by the plant manager on the way. They see the kanban boards that have been refreshed for their visit. They never walk an unannounced shift. They never linger long enough to be told something inconvenient. They never sit through a hansei session where a current failure is named and a current commitment is made in front of them. If the senior leader of your operation does not have a regular, unscheduled, unbriefed presence on the floor, you do not have “genchi genbutsu.” You have ceremonial visitation.

The test: ask the supervisor of any line, in private, when the last unannounced executive presence was. Not a tour. Not a briefing. A presence. If the answer is more than ninety days, the substrate is not there. The executives have been managing the reports.

Run these three signs against any improvement program in your career. The pattern is consistent. Artifacts present, system absent, program dissolves within the tenure of the next senior change. The improvement effort survives as a slide in next year’s strategy deck. The operation reverts. The next consulting engagement is already being scoped.

What this means for your operation now.

This is what the publication exists to name. The Lean industry sells implementation. What you need is diagnosis. Implementation without diagnosis gives you another decorated operation. Diagnosis without implementation gives you the option of building a real system this time, if you have the substrate. Most organizations do not have the substrate, and the honest answer is that they should stop calling their improvement work Lean. They should call it what it is. Cost extraction with Toyota wallpaper.

The harder question and the one worth carrying out of this piece, is whether your organization has the substrate to support a real system. Not the tools. The substrate. Long-term commitment to the people on the floor. Supplier relationships you intend to keep through the next downturn. Senior leaders who actually go to the work without being briefed. Standards that change because the floor changed, not because the consultant left a template.

If you have the substrate, you can build a system. Slowly. Over years. The tools will look familiar to anyone who has read Liker or Ohno. The compounding will not. The compounding is what no copier ever got, because no copier ever paid for the substrate.

If you do not have the substrate, the most honest move is to stop pretending. The pretense costs more than the recognition. The pretense burns a generation of operators on a program that cannot work, and the operators who lived through it carry a quiet cynicism about improvement work that takes another decade to repair.

Toyota built something rare. The rarity was not the tools. The tools have been in print for forty years. The rarity was the willingness to build the substrate, hold it for six decades, and let the system compound at the pace the substrate allowed.

The kanban board on your wall is a question. The system is whether you can read the answer.

About PIOL

We translate complexity into signal, helping leaders see what is changing, what it means, and where attention is required before risk compounds or opportunity passes.

If you’re new to PIOL – Practical Intelligence for Organizational Leadership, you can follow for free to get future issues or paid to expanded views for actionable guidance.

Subscribe now

Leave a comment

Share Pi of Leadership by PIOL

Visit piol.ai to learn more.