Walk the head of a line in a manufacturing facility and the board tends to present the same picture: cards arranged by color code, slots populated, ownership names legible, the whole surface maintained and orderly. Ask on the same floor where work orders actually originate, and the answer does not point to the board; it points to the schedule produced in the weekly planning meeting and to the demand forecast feeding that schedule. Cards are circulating, yet the signal triggering production is not the card. The card has become a recording instrument that follows a production decision already taken elsewhere. This is a conversion rarely noticed on the floor, precisely because every visible element of the system remains exactly where it was installed.
A second pattern is observable on the same floor. Asked when the card count was last changed and on what basis, the answer generally reaches back to the commissioning period of the line, a time when the product mix was appreciably narrower than it is today; the person who performed that calibration is frequently still in the plant and knows the right number from experience, but the knowledge sits in one individual's memory rather than in a calculation record. A third pattern concerns exception flow: when a line-side item runs dry, the mechanism that engages is not the board but a message to the warehouse supervisor or a call placed to the supplier, and over time this exception normalizes to the point where it is no longer treated as an exception and no one counts it.
The mechanism operating here is kanban, and its logic is elementary: authority to produce or replenish arises from the actual consumption of the next station, not from a forecast. The card is the document conveying that authority, while the number of cards in circulation is the ceiling physically constraining work-in-process and stock within the system. The correct level of that ceiling emerges from three variables in combination — consumption rate per unit of time, container-to-container replenishment lead time, and a safety allowance absorbing the variability in both — divided by container size to yield a card quantity. The elegance of the system resides here, in reducing a complex planning computation to a physical constraint. So does its fragility, since all three variables drift over time while the card quantity, on its own, can remain fixed.
That this structure works under particular conditions is no accident. Where the mix is narrow, consumption steady, supplier lead time predictable, and quality stable, the card ceiling reduces planning burden to nearly nothing; and the system carries a second benefit, often more valuable than the first, in that as the ceiling is lowered the line is deprived of the buffer stock that would otherwise conceal its problems, so breakdowns, quality drift, and long changeover times surface by stopping the line rather than hiding behind inventory. Viewed this way, kanban is as much a diagnostic instrument as a replenishment method, and in mature applications the card count is not a target but a deliberately compressed diagnostic parameter.
When conditions change, the same mechanism begins to run in the opposite direction. As the product mix broadens, each variant demands its own card set, and the total number of cards in circulation grows item by item without anyone taking a conscious decision to expand it. As lead times extend — particularly on single-sourced overseas items — the line runs dry unless the safety allowance is enlarged proportionally, and once enlarged, the system ceases to be pull at all and becomes a minimum-stock policy expressed in card form. Where engineering change velocity is high, obsolete cards for a revised part continue circulating, and unusable inventory becomes invisible at exactly the point where the system promised visibility. The shortcut itself is not the error; the error is sustaining the shortcut after the condition producing it has disappeared.
The financial trace of this degradation is rarely legible in total inventory, since the total usually stays within a defensible band. It appears instead in composition and in the item-level dispersion of turnover: a handful of fast-moving items carry the average while items whose card sets were never revised turn a full order of magnitude more slowly, and that tail eventually reaches the income statement as an obsolescence provision. A second trace sits on the cost side, where off-system expedites accumulate in premium freight, split shipments, and overtime, though these lines are typically consolidated under logistics or labor expense and are therefore never traced back to the performance of the replenishment system. A third trace lies in line downtime hours, where the recorded cause is generally material shortage rather than stale card calibration.
Managed separately these three traces appear tractable, yet placed before an acquirer or a lender they consolidate under a single heading: operational predictability. The question raised at a review table is not whether the plant operates kanban, since the board has already been seen; the question is on what cadence, with which inputs, and under whose authority the card count is recalculated, and to the extent that the answer rests on the experience of one senior employee rather than on a calculation record, the finding takes shape as person-dependent operational performance. The consequence within the transaction structure is predictable enough: a more conservative normalization of working capital, earn-out triggers conditioned on key-personnel retention, or a broader representation and warranty package covering inventory valuation.
Extending the system to suppliers deepens the layer further. Supplier kanban, consignment stock, vendor-managed inventory, and line-side feeding arrangements resemble one another operationally while diverging legally, and the critical distinctions concern the moment at which title passes, which party bears the risk on unconsumed stock, what each side owes when minimum or maximum bands are breached, and where insurance coverage begins. On the floor these arrangements frequently rest on a one-page work instruction annexed to a framework agreement, and because the instruction is drafted as an operational document, it does not establish risk allocation with any precision. Where a capacity contraction or a payment strain emerges on the supplier side, that ambiguity surfaces at precisely the moment when negotiating leverage is at its lowest.
What neutralizes this tendency is not individual attentiveness but a governance design composed of four separable components. The first is calibration cadence and ownership: recalculation of the card count is placed on a calendar, the three inputs to the calculation — consumption rate, replenishment lead time, observed variability — are recorded, and authority to change the count is assigned to one named role. The second is the exception log: every expedite executed outside the kanban is counted together with its cause and item, since an unmeasured exception eventually becomes the system itself. The third is written stop authority: where the operator holds authority to halt the line when the ceiling is breached or an item runs dry, the system functions as a diagnostic instrument, whereas authority moved upward quietly restores buffer stock. The fourth is the contractual mirror, verifying that the physical arrangement at the line and the risk allocation in the supply agreement state the same thing.
The intervention BEIREK makes on this line in capital-intensive facility projects is positioned at the threshold between commissioning and steady-state operation, since the initial calibration of the card set is performed precisely at that threshold and is seldom revisited afterward. The mechanism installed rests on three records: a calibration record holding the inputs to the calculation and the rationale for each change under a date stamp, thereby closing the gap between the person who knows the right card count and the institution able to read the record; an exception record accumulating every off-kanban pull by item and cause, rendering the true performance of the system measurable; and a contract-operations reconciliation record comparing the arrangement actually in force at the line against the corresponding clause of the supply agreement. The cadence running on top of these records fixes the card set as a standing agenda item in the monthly operations review.
The principal return on such an arrangement is not an improvement in inventory levels; the return is the conversion of operational performance from person-held experience into an institutional process, and what determines the valuation of an asset is frequently not performance itself but the ability to demonstrate that the performance is repeatable independently of the founder and of key personnel. The presence of a board is not evidence of the presence of a system; the evidence is the ability to show, in writing, when the card count last changed and on what grounds.