And what separates deviation from discipline
There is a question that doesn’t get asked enough in organisations that take Lean seriously.
If standard work is the foundation of everything — the baseline from which all improvement begins, the documented proof that a process works, the thing we train people on, audit against, and hold up as operational truth — then why do the best operators quietly deviate from it?
Not because they’re careless.
Not because they haven’t been trained.
Not because they don’t understand what it says.
They deviate because they know something the standard doesn’t.
And that gap — between what the standard says and what the expert knows — is one of the most important leadership problems in manufacturing. Get it wrong in one direction and you get a rigid, compliance-driven culture that performs adequately on audit day and slowly degrades the rest of the time. Get it wrong in the other direction and you get operational chaos dressed up as empowerment.
Get it right, and you have the engine of genuine continuous improvement.
This post is about that gap. It’s about why standard work fails, why it gets broken, and why the organisations that treat that breakage as a problem are missing the most valuable signal in their operations.
1. What Standard Work Is Actually For
Let’s establish something that often gets lost in the implementation.
Standard work is not a rulebook.
It is a hypothesis.
It says: based on everything we currently know, this is the best way to perform this task, in this sequence, at this pace, to produce this outcome. It is the crystallised knowledge of everyone who has studied, improved, and stabilised this process to date.
That framing matters enormously, because it changes the relationship between the person doing the work and the document on the wall.
If standard work is a rulebook, deviating from it is a disciplinary matter. The role of the leader is to enforce compliance. The role of the operator is to follow instructions. Improvement happens when management decides it should.
If standard work is a hypothesis, deviating from it is a data point. It either means the standard is wrong, the conditions have changed, or the person doing the work knows something the standard doesn’t. Any of those is worth understanding.
Toyota, which invented the thing, understood this from the beginning. The standard exists to be improved. The improvement comes from the people closest to the work. Without a standard, you cannot improve deliberately — you’re just changing things and hoping. With a standard, every deviation is a potential lesson.
Most Western implementations kept the document and lost the philosophy.
2. The Three Reasons Experts Deviate
When a skilled, experienced operator doesn’t follow standard work, one of three things is happening.
The standard is out of date.
Processes change. Materials change. Tooling wears and gets replaced. Equipment is reconfigured. Customer requirements shift. The engineering change that came through three months ago updated the product but nobody updated the standard. The new consumable has slightly different properties and requires a different handling technique. The machine that was replaced last quarter runs at a marginally different speed.
Standards decay. Not dramatically, not all at once, but incrementally — through a thousand small changes that individually don’t warrant a formal update and collectively make the documented standard increasingly disconnected from operational reality.
When experienced operators deviate in these circumstances, they’re not breaking the standard. They’re running ahead of it. The standard has already broken — it just hasn’t been updated yet.
The standard was always incomplete.
Documentation captures what is observable and describable. It rarely captures everything that matters.
The tacit knowledge that makes an expert genuinely expert is, almost by definition, hard to write down. The feel of a material at the correct tension. The sound of a machine running at the edge of tolerance. The early visual indicator of a quality issue that will manifest two steps downstream. The sequencing adjustment that prevents a micro-jam that doesn’t happen often enough to have made it into the standard but that the expert has learned to anticipate.
This knowledge exists. It is real and it produces real results. And it almost never makes it into the standard work document, because the person who wrote the standard work document either didn’t know it, or couldn’t articulate it, or ran out of time on the improvement workshop.
When experts deviate because of tacit knowledge, they are adding value the standard hasn’t yet found a way to contain.
The conditions are genuinely different.
Standard work is designed for standard conditions. It assumes a particular material lot, a particular ambient temperature, a particular equipment state. When those conditions shift outside the envelope the standard was designed for, applying the standard blindly can produce worse outcomes than making a considered adjustment.
The experienced operator knows this. They’ve seen what happens when you run the standard in off-nominal conditions. They’ve learned to recognise those conditions and adapt accordingly.
This isn’t a failure of discipline. It’s exactly the contextual judgement that experienced people are hired for.
The problem — and this is important — is that in most organisations, none of these three categories of deviation is formally distinguished from each other, or from simple non-compliance. All deviation looks the same on the audit sheet. All deviation gets the same response: retrain, remind, reinforce.
And so the signal is lost.
3. What Compliance Culture Actually Produces
There is a version of standard work deployment that is very clean.
The standards are documented. They are comprehensive. They cover every process step and every product variant. They are posted at every workstation, laminated, current. Operators are trained against them at induction and annually thereafter. Audits confirm compliance. Non-conformance is recorded and followed up. Compliance rates are reported in the weekly operations review.
Everything looks right.
And yet something feels stuck. The improvement programme produces events but not results. Suggestions come from management, not operators. Problems get solved and then quietly come back. The factory performs to a level but doesn’t compound. Knowledge walks out the door every time a long-service operator retires.
This is what compliance culture produces.
When standard work becomes primarily an enforcement mechanism, several things happen simultaneously.
Operators learn to perform for the audit rather than perform for the outcome. They follow the standard when they know they’re being watched and adapt it when they’re not. The resulting process has two modes — observed and actual — and they diverge over time.
Tacit knowledge stops surfacing. If suggesting that the standard might be wrong is unwelcome, people stop suggesting it. They keep their knowledge to themselves, apply it quietly, and take it with them when they leave.
Improvement becomes top-down by default. Because the operators aren’t in the loop on standard development, and because deviating from the standard is treated as a problem rather than a signal, all improvement initiative migrates upward. Engineers and improvement specialists drive the change. Operators implement it. The loop that should connect shopfloor insight to documented practice closes at the management layer instead.
The standard becomes a ceiling.
Performance converges on what the document describes and then stops. Because improving beyond the standard requires deviating from it, and deviation is not welcome.
That is the paradox.
The very mechanism designed to protect process performance becomes the thing that limits it.
4. The Other Failure Mode
Fairness demands that we also examine what happens at the other extreme.
Because there is a version of empowerment culture that is equally broken, just less obviously so.
In this version, standard work exists but is not enforced. Operators are trusted to do what they think is right. There is a strong culture of experience and craft. Improvement is organic — people try things, share what works, adjust informally. Leadership prides itself on not being prescriptive.
It feels healthy. It feels respectful. It is, in many respects, the polar opposite of the rigid compliance culture described above.
But it has a different set of problems.
Variability is endemic. Because every operator brings their own interpretation, the same process produces different outcomes depending on who is running it. The experienced operator produces consistently excellent results. The operator eighteen months into the role produces results that are adequate but not the same. The new starter is essentially finding their own way. Quality variation, cycle time variation, and setup time variation are treated as inherent rather than designed.
Knowledge is not transferable. If the best operator in the building is the only person who knows how to run a particular product consistently, that knowledge is a single point of failure. When they leave, retire, or move to another area, the capability goes with them. The organisation hasn’t captured it. It has just relied on it.
Improvement is not cumulative. Because there is no documented baseline, it is genuinely difficult to tell whether a change is an improvement. Today’s practice reflects accumulated tribal knowledge, informed guesswork, and habit in roughly equal measure. Knowing which is which is nearly impossible without the reference point that documented standard work provides.
And perhaps most critically: without a standard, you cannot distinguish between someone who is performing at a genuinely higher level and someone who is taking shortcuts that produce acceptable short-term results at the cost of long-term stability.
Both extremes — rigid compliance and unstructured autonomy — fail in different ways.
The answer sits between them. But it’s not a compromise. It’s a completely different design.
5. The Living Standard
Toyota’s concept of standard work was always dynamic. The standard is the current best-known method. The improvement cycle — observe, analyse, experiment, update — is what advances it. The standard is both the starting point of improvement and the product of it.
The phrase that captures this precisely is: follow the standard until you have a better one. Then make the better one the standard.
That sounds obvious. In practice, it requires several things that most organisations have not designed into their operating system.
It requires psychological safety at process level.
An operator who has found a better way must feel confident that surfacing it will lead to an improvement conversation, not a compliance conversation. If the response to “I think I’ve found a better method” is “why weren’t you following the standard?” then the surfacing stops. The organisation has inadvertently created a system that punishes the signal it most needs.
This is not about being soft. It is about being precise. The question should not be “why did you deviate?” The question should be “what did you learn?”
It requires leaders who can distinguish deviation types.
Not all deviation is equal. The leader who treats an experienced operator’s adaptive response to off-nominal conditions the same way they treat a new starter skipping a safety check is making an analytical error, not just a cultural one.
Great team leaders and supervisors are skilled at reading the difference. They know which deviations represent non-compliance, which represent tacit knowledge, and which represent an out-of-date standard. They treat each one differently, because they are different. That skill is learnable. It is also rarely trained.
It requires a formal pathway from deviation to update.
This is the mechanical piece that most systems are missing.
If an operator identifies that the standard is wrong, what happens next? In most organisations, the answer is: they raise it with their team leader, who raises it with engineering, who schedules it for a future improvement event, where it may or may not make it onto the agenda, and where the resulting update may or may not get formally documented before the event closes and the team disperses.
That pathway is too long, too uncertain, and too dependent on the right people being available at the right moments.
The organisations that do this well have short, formal, predictable loops. The deviation is captured. It is categorised. If it represents a genuine improvement, it is assessed quickly — not in three months at the next kaizen event, but within a defined short cycle. If it is validated, the standard is updated. The person who raised it sees the outcome. The loop closes visibly.
That visibility matters more than people realise. When operators see that their observations lead to changes in the standard, they raise more observations. When they see that observations disappear into a process and nothing changes, they stop. The feedback loop is not just mechanical — it is motivational.
6. What Great Leaders Actually Do
Here is where the post title finds its proper meaning.
The best leaders don’t break standard work carelessly.
They break it deliberately, visibly, and instructively — and then they fix the standard.
They use their own deviation as a teaching moment. They say, out loud, in front of the team: “I’m doing this differently from the standard, and here’s why. If this approach produces better results, we should update the standard. Watch the outcome with me.”
That behaviour does several things at once.
It models the correct relationship with the standard — not compliance, not disregard, but informed and accountable engagement.
It creates psychological safety for others to do the same. If the leader can say “the standard might be wrong,” the operator can say it too.
It demonstrates that the standard is a living document rather than a fixed rule. It can be changed. It should be changed. Changing it is not a sign that the previous standard was a failure — it is a sign that the process is learning.
And it anchors the deviation to an outcome. The leader is not deviating randomly. They are running a deliberate experiment with a hypothesis and an expected result. That is the scientific method applied to process improvement. It is exactly what Lean was always meant to produce.
The leaders who do this well are not breaking standard work. They are practising it at its highest level.
7. The Knowledge Capture Problem
There is a crisis unfolding quietly in most manufacturing businesses.
The generation of operators and technicians who built deep process knowledge over twenty and thirty year careers are retiring. They are taking with them knowledge that has never been written down, that exists only in their hands and their pattern recognition and their accumulated instinct for the edge cases that matter.
Standard work, properly designed and maintained, is the mechanism for capturing that knowledge before it walks out the door.
But it only works if the culture allows tacit knowledge to surface. If operators trust that what they know is valued. If there is a pathway from “I do this differently and it works” to “let me show you and we’ll capture it properly.”
The businesses that are best positioned for the decade ahead are the ones investing now in capturing that knowledge systematically — not just documenting current practice, but actively eliciting the deviations, understanding why they work, and incorporating them into the standard.
This is not a documentation project.
It is a knowledge strategy.
And it starts with creating the conditions under which people feel safe enough to share what they know, even when — especially when — what they know contradicts what the current standard says.
8. The Audit Problem
A word on audits, because they are where this philosophy either comes to life or dies.
Most standard work audits are structured as compliance checks. The auditor observes the process. They compare what they see to what the standard says. Any divergence is recorded as a finding. The finding is followed up. Compliance is tracked over time as a percentage.
This is not a bad thing to do. It is an incomplete thing to do.
The compliance audit answers one question: are people following the standard?
It doesn’t answer the more important question: is the standard still right?
A mature approach to standard work auditing distinguishes between deviation that indicates a compliance problem and deviation that indicates a standard problem. It treats the observation phase of the audit not just as a check against the document but as an investigation of current practice.
The auditor asks: is there anything in how you’re doing this that isn’t in the standard? What do you know about this process that the standard doesn’t capture? Are there conditions under which you’d do this differently?
Those questions require trust. They require the auditor to be seen as a partner in improving the standard rather than an inspector checking for violations. That is a cultural shift. It is also a significant capability shift.
The audit becomes a knowledge-gathering mechanism rather than a compliance-scoring mechanism.
Performance on that audit improves differently. Not because compliance rates go up — though they do — but because the standard improves, and with it the process it describes.
9. Standard Work for Leaders
There is an uncomfortable dimension to this conversation that rarely gets addressed directly.
Standard work is almost exclusively applied to operational and technical processes. It governs how a machine is set up, how a component is assembled, how a changeover is executed, how a quality check is performed.
It is very rarely applied to leadership work.
And yet the activities that most determine operational performance — the daily team meeting, the tier review, the problem-solving conversation, the improvement coaching session — are leadership processes. They have inputs, outputs, and methods. They can be standardised. They can be improved. They exhibit exactly the same variation as any operational process, with exactly the same consequences.
When the daily management meeting is run differently by every team leader in the building — different duration, different content, different tone, different decision quality — the result is the same as running a production process without standard work. Consistent outcomes are impossible. Improvement is random. Capability doesn’t compound.
The businesses that genuinely embed standard work as a leadership philosophy rather than a shop floor tool standardise their management processes with the same rigour they apply to their operational ones. Leader standard work — the documented daily routines, observation methods, escalation protocols, and coaching behaviours that define how leaders operate — is the direct analogue of operator standard work.
And the same paradox applies.
The best leaders don’t follow their own standard work blindly. They follow it until they find a better way. Then they update the standard, share what they’ve learned, and raise the baseline for every leader in the system.
That cycle is what a learning organisation actually looks like from the inside.
10. The Cultural Equation
Everything in this post points to one underlying condition.
The standard work paradox — the fact that the best people break the standards while the least experienced rigidly follow them — only becomes a problem when the culture doesn’t have a mechanism for closing the gap.
In a healthy culture, the expert deviates and the organisation learns. The deviation is captured. The standard is improved. The improvement is shared. The baseline rises. Everyone benefits from what the expert knows.
In an unhealthy culture, the expert deviates and the organisation disciplines. The deviation is suppressed. The knowledge stays private. The standard remains wrong. And the gap between what the standard describes and what the process actually requires grows quietly wider.
The difference is not technical. The tools are the same. The format of the standard work document is the same. The audit schedule is the same.
The difference is whether the leader treats every deviation as a compliance failure or as a question worth asking.
That question — why is this person doing it differently? — is one of the most powerful diagnostic tools available to an operational leader. Follow it with genuine curiosity, not defensive scrutiny, and you will find:
Out-of-date documentation that needs updating. Tacit knowledge that needs capturing. Process conditions that need stabilising. Training gaps that need addressing. And occasionally, compliance failures that need correcting.
All of those are useful. None of them surface if the question is never asked.
11. The Compound Effect
Here is what changes when an organisation gets this right.
Not immediately. Not dramatically. But sustainably, over a period of years.
The standard improves continuously. Not because a kaizen event happened, but because every deviation is treated as a data point and every valid data point updates the baseline. The standard this year is better than last year’s standard. Next year’s will be better still. The improvement is small in any given cycle and significant over a decade.
Tacit knowledge becomes organisational knowledge. The expertise that used to live in individual heads gradually migrates into documented practice. New starters reach competence faster. Experienced operators can transfer to new areas without the organisation losing their knowledge. Retirement is planned for, not recovered from.
Variability reduces. Not because operators are told to comply more firmly, but because the standard more accurately reflects what the best practice actually is. Following it produces better results than deviating from it, which is the only sustainable basis for genuine compliance.
Operators become improvement partners. Because they trust that their knowledge will be heard, they share it. Because they see their observations lead to real changes, they observe more carefully. The improvement capability of the organisation grows — not concentrated in a central function, but distributed across every person doing the work.
And leadership attention shifts. When the standard work system is working as designed, the exception rate falls. The firefighting reduces. The audit findings become more substantive and less frequent. The improvement conversation gets more interesting.
Leaders spend less time enforcing standards and more time improving them.
That is the shift.
Final Thought
The standard work paradox resolves when you reframe what standard work is for.
It is not a compliance instrument.
It is a knowledge management system.
Its job is to capture what the organisation currently knows, make that knowledge accessible to everyone, and create the conditions under which that knowledge advances.
The best operators in your business are already advancing it. They’re doing it quietly, locally, and without formal recognition. Every adapted technique, every workaround that actually works, every tacit adjustment that separates the expert from the average — that is potential organisational knowledge, waiting to be captured.
The only question is whether your system is designed to receive it.
Three questions to sit with.
When did your standard work last change because an operator found a better way — not because an engineer designed one?
Do the best people in your operation follow the standard, or do they follow something better that nobody has written down yet?
And if someone in your team broke a standard today and it turned out they were right — what would happen next?
The answer to that last question tells you almost everything you need to know about whether your standard work system is a ceiling or a ladder.
adam
Leave a comment