The Promise of the Gemba
The concept is beautifully simple. Genchi genbutsu — go and
see for yourself. Taiichi Ohno famously drew a chalk circle on the
factory floor, stood a young engineer inside it, and told him to watch
until he understood the waste. No dashboards. No summary reports. No
filtered data through three layers of management. Just the real work,
happening in real time, observed by the people responsible for improving
it.
The promise of the gemba walk is that leaders and engineers leave
their offices, stand where value is actually created, and see what is
really happening. Not what the metrics say is happening. Not what the
shift report claims happened yesterday. Not what the standard work
document says should be happening. What is actually happening,
right now, on the shop floor, in the process, at the point of
production.
It is the single most powerful — and most universally corrupted —
practice in the entire lean manufacturing toolkit.
What Goes Wrong
Here is what actually happens in most organizations that adopt gemba
walks.
The Vice President of Operations announces that every manager will do
a weekly gemba walk. A schedule is created. A route is mapped through
the plant. A checklist is distributed: “Are workstations clean? Are
operators wearing PPE? Are 5S standards maintained? Are visual
management boards updated? Are tools in their assigned locations?”
Every Wednesday morning, a group of managers — sometimes eight or ten
of them — walks the designated route in a cluster. They spend three to
five minutes at each station. They look at the visual management board.
They glance at the work area. They check items off the list. They nod at
operators who freeze up because ten managers are staring at them. They
take notes on their clipboards. After forty-five minutes, they return to
the conference room, review findings, assign action items, and file the
report.
This is not a gemba walk. This is an audit performed outdoors.
The distinction matters because the two activities have completely
different purposes. An audit verifies compliance against a standard. A
gemba walk builds understanding of the actual process. When you turn a
gemba walk into an audit, you get the worst of both worlds: you lose the
depth of a real audit (because you’re moving too fast and checking
superficial items) and you lose the learning of a real gemba walk
(because you’re focused on checking boxes instead of observing
flow).
The Seven Failure Modes
After twenty-five years of visiting manufacturing plants across three
continents, I have observed seven recurring failure modes that transform
gemba walks from a learning practice into a performative ritual.
1. The Entourage Problem
The single most common failure. A gemba walk is not a group tour.
Ohno sent one engineer to stand in one circle. The purpose was
sustained, individual observation — the kind of focused attention that
is physically impossible when you are part of a delegation of nine
managers walking in a formation, each trying to look smart in front of
the others.
When you walk in a group, several things happen. Operators behave
differently — they become self-conscious, slow down, follow procedures
more carefully than normal, and hide the shortcuts they usually take.
You see a performance, not the process. The group dynamic also shifts
the cognitive mode: instead of observing, participants are managing
impressions — asking questions that demonstrate their knowledge,
pointing out problems that show their attentiveness, deferring to the
senior person’s observations rather than forming their own.
A real gemba walk is done alone or with one other person. Two people
maximum. If your VP insists on joining, fine — but one person leads, one
person observes, and they debrief afterward. The moment you have more
than three people, it is no longer a gemba walk. It is a procession.
2. The Checklist Substitution
This is the failure mode that kills learning most efficiently. A
checklist feels productive. It feels structured. It feels like you are
doing real work. But a checklist converts observation into verification,
and those are fundamentally different cognitive activities.
When you verify, you ask: “Does this match the standard?” The answer
is binary — yes or no — and the thinking stops there.
When you observe, you ask: “What is actually happening here, and
why?” The answer is open-ended, contextual, and leads to further
questions. Why is that operator reaching across the workstation? Why is
there a small pile of parts between operations three and four? Why does
the andon light keep flashing yellow even though the shift supervisor
says everything is fine? Why does the standard work say the cycle time
is forty-five seconds when the operator is clearly taking sixty?
Verification closes questions. Observation opens them. The checklist
is designed to close questions efficiently, which is precisely why it
prevents the discoveries that gemba walks are supposed to produce.
This is not an argument against checklists in general. Checklists are
extraordinary tools for ensuring consistency in repetitive tasks —
surgical safety checklists, pre-flight checklists, preventive
maintenance checklists. But gemba walks are not repetitive tasks. They
are exploratory observations of a complex, dynamic system. Applying a
checklist to a gemba walk is like applying a ruler to an ocean — you
measure something, but you measure the wrong thing.
3. The Speed Trap
Most gemba walks I have observed last between thirty and sixty
minutes and cover the entire plant. That is a walking speed of
approximately one station per three to five minutes. At that pace, you
cannot observe anything meaningful. You can verify that the area looks
clean. You can confirm that the visual management board has been
updated. You can check that the operator is present and appears to be
working.
But you cannot observe flow. Flow takes time to see. You need to
stand in one place long enough to watch at least ten complete work
cycles — ideally twenty — to see the variation, the micro-stoppages, the
rework, the waiting, the repositioning that operators do so smoothly it
looks intentional but is actually compensation for a poorly designed
workstation.
Ohno’s engineer stood in that chalk circle for hours. Not minutes.
Hours. He watched the same operation repeatedly until the waste became
visible to him. The waste was always there — it just required sustained
attention to see it.
If you walk through your plant in forty-five minutes, you have not
done a gemba walk. You have taken a walk. There is nothing wrong with
taking a walk, but do not confuse it with the practice that built the
Toyota Production System.
4. The Hierarchical Filter
When a senior leader visits the shop floor, the visit is almost
always pre-announced. The shift supervisor prepares the area. The team
leader briefs the operators. The visual management boards are updated
with fresh data. The accumulation of partially completed work is cleared
away. Everything that looks bad is made to look acceptable.
This is not deception. It is human nature. When the boss visits your
workspace, you clean up. Everyone does this — you did it when your own
boss visited, and your boss did it when their boss visited. The problem
is that a pre-announced gemba walk is a gemba walk of a stage set, not
of the real workplace.
The hierarchical filter operates on multiple levels. The VP sees what
the plant manager wants them to see. The plant manager sees what the
shift supervisor wants them to see. The shift supervisor sees what the
team leader wants them to see. At each level, information is filtered,
problems are minimized, and the picture becomes progressively more
reassuring and progressively less accurate.
The only way to defeat the hierarchical filter is to go alone, go
unannounced, go at different times of day, and go to the same place
repeatedly over weeks and months. Operators need to see that your
presence on the shop floor is normal, not an event. Only when the gemba
walk becomes routine — as routine as checking email — will the shop
floor stop performing for you and start showing you the truth.
5. The Solution Reflex
This failure mode is particularly common among engineers and managers
who are genuinely trying to help. They see a problem on the shop floor —
an operator struggling with a fixture, a process step that takes too
long, a workstation with poor ergonomics — and they immediately propose
a solution. Sometimes they propose it on the spot. Sometimes they write
it up as an action item. Either way, the solution reflex is activated
within seconds of seeing the problem.
The solution reflex is dangerous because it terminates observation
prematurely. The first explanation for a problem is almost never the
real root cause. The operator is struggling with the fixture — why? Is
the fixture worn? Is it the wrong fixture for this part variant? Was the
part loaded incorrectly upstream? Is the operator new and undertrained?
Is the fixture correct but the clamping force has drifted because
maintenance hasn’t checked it in eight months?
Each of these explanations leads to a completely different corrective
action. If you jump to a solution based on the first five seconds of
observation, you will implement a fix that addresses the symptom and
leaves the root cause untouched. The problem will return — maybe in a
different form, maybe in the same form — and the operator will learn
that when they show a problem to a manager, the response is a quick fix
that doesn’t actually solve anything.
The antidote to the solution reflex is discipline: when you see a
problem during a gemba walk, your job is to understand it, not to solve
it. Ask questions. Follow the process upstream. Talk to the operator.
Gather data. Then leave, think about it, and come back with a considered
response. The speed of your response is far less important than the
quality of your understanding.
6. The Metrics Disconnect
Many organizations have adopted the practice of displaying
performance metrics at each workstation — the visual management board,
the production tracking chart, the OEE dashboard. These are valuable
tools. But during a gemba walk, they become a trap.
Leaders walk up to the board, read the numbers, ask a question or two
about the trend, and move on. They have engaged with the representation
of the process rather than the process itself. It is the difference
between reading a map and walking the terrain. The map shows you the
route; the terrain shows you the rocks, the mud, the steep inclines, and
the washed-out bridges.
Metrics are a model of reality, not reality itself. They are always
lagging, always aggregated, always smoothed. An OEE number of
seventy-two percent tells you nothing about the eleven micro-stoppages
that happened during the shift, each lasting forty seconds, each
invisible in the data because the system rounds down. But if you stand
at the machine for twenty minutes and watch, you will see those
stoppages. You will see the operator restart the machine, clear the
minor jam, and resume production — all within forty seconds, all so
smoothly that it looks normal. It is not normal. It is a slow leak in
your production capacity that the metrics cannot detect but your eyes
can.
The metrics board is a starting point for a gemba walk, not the
destination. Read the board, note the numbers, then turn around and
watch the actual process. The gap between what the board says and what
the floor shows is where the learning lives.
7. The Event Mentality
The deepest and most pervasive failure mode is treating the gemba
walk as an event rather than a habit. When gemba walks are scheduled —
every Wednesday at 9 AM, for instance — they become one more item on the
calendar, sandwiched between the production meeting and the cost review.
They are something to get through, something to complete, something to
check off.
Toyota managers did not schedule gemba walks. They lived on the shop
floor. Ohno’s office was on the production floor — not adjacent to it,
not overlooking it, on it. The gemba was not a place you
visited; it was the place you worked. The idea of scheduling a recurring
event to go see the place where value was created would have struck Ohno
as absurd, like scheduling a recurring event to breathe.
This does not mean that modern managers — who have meetings, reports,
emails, and cross-functional responsibilities — can realistically
relocate their offices to the shop floor. But it does mean that the goal
is frequency and normality, not formality and ceremony. Five ten-minute
walks per day, unannounced, at varying times, to different areas, are
worth more than one ninety-minute scheduled walk per week. The total
time is similar. The learning is incomparable.
What a Real Gemba Walk Looks
Like
Having catalogued the failure modes, let me describe what a genuine
gemba walk looks like in practice. Not the idealized version from a
textbook — the real version, as practiced by the few organizations and
individuals who do it well.
You go alone. You pick one area — one workstation, one operation, one
process segment — and you stay there. For at least twenty minutes. You
stand close enough to see the work clearly but far enough that you do
not make the operator uncomfortable. You watch complete work cycles. You
count. You time. You note the variation between cycles. You look at the
material flow — where do parts come from, where do they go, what happens
in between?
You ask questions — but not leading questions, not judgmental
questions. “Can you walk me through what you’re doing?” “How often does
that happen?” “What do you do when that doesn’t work?” “What would make
this easier?” These are questions born of curiosity, not
interrogation.
You take notes. Not on a checklist — on whatever you observe that
surprises you, that contradicts what you expected, that doesn’t match
the standard, or that reveals something the metrics didn’t show. You
look for the small things: the micro-adjustments, the repositioning, the
brief hesitations, the parts set aside and then picked up again.
You leave without proposing solutions. You go back to your desk with
raw observations, and you think about what you saw. You cross-reference
with data. You talk to the process engineer. You form a hypothesis. And
then you go back — the next day, or the day after — and observe again to
test it.
This is the practice. It is slow, it is unglamorous, and it produces
no deliverable that can be filed in a report. But it is the only way to
see what is actually happening, and seeing what is actually happening is
the prerequisite for improving anything.
The Leadership Trap
There is one more failure mode that deserves special attention
because it is the hardest to fix: leaders who do gemba walks for the
wrong reason.
Some leaders walk the floor to be seen walking the floor. The gemba
walk becomes a visibility exercise — proof that the leader is hands-on,
connected to operations, in touch with the front line. The walk is
performative. The leader moves through the plant with a entourage, stops
at key stations for photo opportunities, asks a few questions that
demonstrate their manufacturing knowledge, and leaves having added zero
understanding to their mental model of the process.
Other leaders walk the floor to solve problems in real time. They see
an issue, gather a team, and initiate a corrective action on the spot.
This feels decisive and effective. It is also the wrong approach.
Real-time problem-solving during a gemba walk turns the walk into a
firefighting exercise and reinforces the expectation that the leader’s
presence means action — which means the team will perform for the leader
rather than show them reality.
The right reason to walk the floor is to learn. To build and
continuously refine a mental model of how the process actually works —
not how it works on paper, not how it worked when it was designed, but
how it works today, in current conditions, with the current people,
materials, and equipment. Every gemba walk should leave you with a
slightly more accurate mental model than you had before. If it doesn’t,
either you weren’t looking or you already know the process better than
anyone alive.
The Path Back
If your organization’s gemba walks have devolved into the failure
modes described above, the path back is not complicated, but it requires
discipline.
Stop scheduling formal walks. Stop distributing checklists. Stop
going in groups. Stop pre-announcing visits. Instead, pick one process.
Go to it tomorrow, alone. Stand there for thirty minutes. Watch. Take
notes on what you see, not what you expected to see. Come back the next
day. Do it again. Repeat for a week.
At the end of the week, you will know more about that process than
anyone in your organization who has not been doing this. You will see
waste that the metrics don’t capture. You will understand constraints
that the production meetings keep discussing without resolving. You will
have questions that the standard work document cannot answer.
That is the beginning of a real gemba practice. Not an event. Not a
program. Not a KPI. Just a habit — the habit of going to see, which is
the foundation of everything else in lean manufacturing and quality
improvement.
Ohno’s chalk circle was not a technique. It was an attitude: the
belief that the truth lives on the shop floor, not in the office, and
that seeing the truth requires the patience to stand still and watch
until it reveals itself.
Every failure mode I have described is, at its root, a way of
avoiding that patience. The checklist replaces patience. The group tour
replaces patience. The speed walk replaces patience. The solution reflex
replaces patience. The metrics board replaces patience. The scheduled
event replaces patience.
And the result is always the same: an organization that believes it
is practicing genchi genbutsu while actually practicing something closer
to its opposite — a comfortable ritual that confirms what leaders
already believe rather than revealing what they need to learn.
Break the ritual. Go alone. Go slowly. Watch. The truth is there,
right where it has always been, waiting for someone patient enough to
see it.
About the Author: Peter Stasko is a Quality
Architect with over twenty-five years of experience in manufacturing
quality management, process improvement, and lean transformation. He has
worked with organizations across automotive, electronics, aerospace, and
medical device industries to build quality systems that deliver
measurable results rather than impressive documentation.
