The Promise That Sold You
the Cards
You went to the seminar. You read the book. You bought the colored
bins, the laminated cards, the magnetic boards. Your team spent three
months mapping every workstation, calculating takt times, and
determining how many units constituted a “reasonable” buffer between
processes. You trained every operator on the rules: when a bin is empty,
send the card upstream. When the card arrives, produce. When there are
no cards, stop. Simple. Elegant. Pull instead of push. Inventory drops,
lead times shrink, problems surface. That was the promise.
Six months later, you have cards pinned to boards that nobody reads,
bins overflowing because “we might need them,” and a production schedule
that looks suspiciously like the MRP push system you were supposed to
replace. The kanban system is technically in place. The kanban
discipline left the building three weeks after the kickoff
event.
This is not a story about kanban failing. Kanban doesn’t fail.
Organizations fail at kanban. And the way they fail follows patterns so
predictable that if you walked into any manufacturing facility claiming
to run a pull system, you could diagnose the gap in under an hour.
How Pull Becomes
Push: The Five-Stage Decay
Stage One: The Pilot That
Worked
Every kanban implementation starts with a pilot cell. Two
workstations, one product family, a consultant who watches like a hawk
for the first six weeks. The pilot works beautifully. Inventory drops by
thirty percent. Lead time compresses. The team feels empowered.
Management sees the numbers and says, “Scale it.”
The pilot worked because the conditions were controlled. One product
family. Short distances. A consultant who corrected violations in real
time. Nobody mentions that the pilot existed in an environment that bore
no resemblance to the complexity of the full production floor. The
success of the pilot becomes the evidence that justifies a rollout that
nobody is ready for.
Stage Two: The Rollout Nobody
Owns
The rollout happens because the pilot proved it works. A
cross-functional team maps the entire value stream. Cards are printed.
Bins are labeled. Training sessions are conducted. And then the team
disbands. The consultant leaves. The internal champion gets reassigned
to a new initiative.
Nobody owns the kanban system on a daily basis. There is no audit
cadence. There is no escalation path when the rules are broken. The
system exists as a collection of artifacts — cards, bins, boards —
without the governance structure that keeps them functioning. Within
weeks, operators begin improvising. A card gets lost, so someone prints
a replacement without removing the old one from circulation. Now there
are two signals for one slot. The system quietly oversupplies.
Stage Three: The
Emergency That Broke the Rules
Then the first real test arrives. A key customer expedited an order.
A supplier shipped late. A machine went down for four hours. The kanban
system said “stop producing” because the downstream buffer was full. The
supervisor looked at the empty schedule for the next two hours, thought
about the customer screaming on the phone, and said, “Run it
anyway.”
That decision was not unreasonable. It was, in fact, the correct
short-term call given the circumstances. But the supervisor did not
restore the kanban balance afterward. The emergency override became the
new normal. Every week has an emergency. Every emergency justifies
running without cards. Within a month, the pull system is a push system
with decorative kanban cards attached to bins that are filled according
to the supervisor’s gut feeling.
Stage Four:
The Metrics That Reward Non-Compliance
Here is where the system calcifies into its broken state. Your
organization measures operator utilization. Your organization measures
daily output. Your organization measures on-time delivery. None of these
metrics reward kanban compliance. In fact, kanban compliance actively
penalizes all three.
When kanban says “stop,” utilization drops. When kanban limits
work-in-process, output may temporarily decrease because downstream
constraints are being respected. When kanban surfaces a quality problem
by stopping production, on-time delivery takes a hit.
Your metrics system is pushing in the opposite direction of your pull
system. Operators and supervisors are rational actors. They optimize for
what is measured. If you measure output and utilization, they will push
work regardless of what the cards say. The kanban system becomes a
compliance exercise — something audited during the monthly lean review
and ignored every other day of the month.
Stage Five: The Fossil Record
At this point, the kanban system is a fossil. It exists in the
organization’s layers as a remnant of a previous initiative. The cards
are still there, yellowed and curled on their hooks. The bins still have
kanban squares painted on the floor. New employees are told, “That’s the
kanban system” during orientation, and then immediately taught how
production actually works by operators who learned the
workarounds from the people before them.
Nobody can explain why the system was put in place, what problem it
was supposed to solve, or what good looks like. The entire apparatus —
the cards, the bins, the boards, the training materials — has become an
archaeological artifact. It is the organizational equivalent of an
appendix: it served a function once, that function is gone, and now it
just sits there, occasionally causing problems that require attention
but providing no value.
The Mathematics of
Discipline
The elegance of kanban lies in its simplicity, but that simplicity
conceals a mathematical discipline that most organizations abandon
without realizing it. A kanban system is a closed-loop control
mechanism. The number of cards in circulation between two workstations
directly determines the maximum work-in-process inventory between them.
This number is calculated based on demand rate, replenishment lead time,
and safety factor. Change any of those three variables without adjusting
the card count, and the system drifts.
Most organizations set the card counts during the initial
implementation and never revisit them. Demand shifts. Lead times change
as suppliers come and go. New products are introduced. Old products are
retired. But the card count stays frozen at the number someone
calculated eighteen months ago based on conditions that no longer
exist.
The result is a system that either overproduces (too many cards for
current demand) or starves downstream operations (too few cards for
increased demand). Both outcomes look like “kanban doesn’t work,” but
the real cause is organizational neglect. The system was never
maintained. The math was done once and then treated as permanent truth
rather than a parameter requiring regular recalculation.
The Card Honor Problem
The single most common failure mode in kanban implementations is also
the most basic: people do not honor the cards. This sounds simple, but
the reasons behind it are structural.
An operator receives a kanban card signaling them to produce a
specific quantity. But the shift supervisor has a daily target that was
set without reference to the kanban system. The materials handler is
evaluated on keeping lines stocked, not on respecting pull signals. The
planner is measured on schedule attainment, which was built from a
forecast that assumed push logic.
Every role in the organization has been given performance
expectations that predate the kanban system. The kanban system was
layered on top of those expectations without removing or modifying them.
So when a kanban signal conflicts with a supervisor’s target, the target
wins. When a kanban limit conflicts with a planner’s schedule, the
schedule wins. The cards are the lowest-authority instruction on the
shop floor, and everyone knows it.
Fixing this requires something far more difficult than printing new
cards or running refresher training. It requires realigning every
performance metric, every daily target, and every escalation path so
that the kanban system is the operating system, not a suggestion. Most
organizations are not willing to do this work. They want the benefits of
pull without the operational discipline that makes pull possible.
Electronic Kanban:
The Expensive Distraction
When the paper cards prove unworkable, many organizations turn to
electronic kanban. Barcode scanners. RFID tags. Digital dashboards. The
reasoning is seductive: if the problem is that cards get lost, ignored,
or mismanaged, then digitizing the cards will solve the problem by
making them impossible to ignore.
It does not work. Electronic kanban solves the card-tracking problem
without solving the discipline problem. A digital signal that says “stop
producing” is ignored just as easily as a paper card that says the same
thing. The difference is that the digital system costs forty thousand
dollars in software licenses, requires an IT person to maintain, and
generates beautiful reports that nobody reads.
The failure mode is identical. Supervisors override the system.
Operators work around the scanners. The daily targets still conflict
with the pull logic. The electronic system adds visibility without
adding authority. Now you can see exactly how non-compliant your
organization is, in real time, on a high-resolution dashboard. This is
not improvement. This is surveillance theater applied to a problem that
requires structural change.
What Working Kanban
Actually Looks Like
A functioning kanban system looks boring, and that is precisely the
point. When you walk a floor with a working pull system, you see
stations that are sometimes idle — and nobody panics. You see operators
who stop when their downstream kanban is full — and nobody yells. You
see supervisors who look at the card board before looking at the
production schedule — and nobody questions it.
You also see a system that is actively maintained. Card counts are
reviewed quarterly. Demand changes trigger recalculations. New product
introductions include kanban planning as part of the launch checklist.
The system has an owner — a real person with a real job title who is
accountable for the integrity of the pull logic and who has the
authority to push back when other functions try to override it.
Most importantly, the metrics are aligned. Operator utilization is
not the primary measure. Daily output targets are derived from the
kanban system, not imposed on it. Escalation paths exist for when the
system signals a problem, and those paths lead to engineering or
maintenance responses, not to mandates to run faster.
The Uncomfortable Truth
The uncomfortable truth about kanban implementations is that they
require organizations to give up control in order to gain control. A
push system gives management the illusion of control — you set the
schedule, you push the work, you measure the output. A pull system
surrenders that illusion. The pace of production is determined by actual
customer demand, filtered through a system of signals that you cannot
override without breaking the system.
Organizations that succeed with kanban are organizations that have
made this philosophical shift. They have accepted that the pull system
knows more about optimal production flow than any planner’s forecast.
They have restructured their metrics to reward system compliance rather
than individual heroics. They have given the kanban cards the highest
authority on the shop floor — higher than the supervisor’s target,
higher than the planner’s schedule, higher than the expediter’s
panic.
Organizations that fail with kanban are organizations that want the
inventory reduction without the operational humility. They want lean
results with push logic. They want the cards on the wall and the
inventory in the aisles. They want the language of pull without the
discipline of actually stopping when the system says stop.
Kanban is not a tool. Kanban is a discipline. The tool is a card and
a bin. The discipline is honoring what the card says, even when every
other signal in your organization is telling you to do something else.
That discipline cannot be purchased, trained, or digitized into
existence. It must be built, maintained, and defended — every day, by
people who understand that the cost of overriding the system once is the
erosion of the system’s authority forever.
Peter Stasko is a Quality Architect with over 25 years of
experience in manufacturing quality management, lean implementation, and
production system design. He has guided organizations through the messy
realities of kanban, Lean, and Six Sigma transformations — and has seen
every stage of the decay described in this article firsthand.
