In a production planning meeting, setup time behaves unlike almost any other number the plant produces: it is not measured, it is declared. Asked how long a die change or a fixture swap will take, the operations side generally answers not with the duration observed last time but with the duration that can be safely committed to this time, and the gap between those two figures settles into the system as an allowance nobody has to name. The same plant will frequently show a setup standard defined three years earlier still sitting at its original level, and on certain lines adjusted somewhat upward, even though the crew has performed that identical changeover several hundred times since, the tooling set has been standardized, and the lifting equipment has been replaced. The learning curve has done its work; the standard has not.
The peculiarity of this number stems from the fact that it performs two incompatible jobs inside the same facility at the same time. It is an input to capacity planning — the weekly schedule, the delivery commitment, and the lead time quoted to the customer are all built on it — while simultaneously functioning as the threshold against which shop-floor performance is assessed, the boundary that, once crossed, requires an explanation. A figure asked to serve as both target and defensive line ceases to be a technical parameter and becomes a negotiated one. Under these conditions, the party with the authority to lower the number has no reason to do so, whereas a request to raise it can always be justified; a single difficult shift will typically be treated as sufficient evidence for a permanent upward revision.
The pattern has a name — setup-time inflation, the progressive detachment of the declared changeover duration from the actual technical content of the work, followed by its upward fixation. The mechanism's core defect lies in a single clock being used to time two different things: the interval between machine stop and the first approved part is not the same quantity as the preparation work genuinely performed within that interval. Embedded in the stopped-machine window are items that never required the machine to be idle at all — waiting for the crane, locating the fixture, waiting for a gauge to free up, waiting for first-article approval to come back from quality. So long as work that could be performed external to the machine continues to be timed inside machine hours, the setup standard records not a technical parameter but the aggregate coordination weakness of the plant.
It is necessary to see where this inflation is functional, because otherwise the intervention begins from the wrong place. The allowance carried inside the declared duration is, in substance, a risk premium, and it is priced by the party that actually bears schedule exposure. A missing fixture, a delayed quality release, or an unexpected dimensional deviation fails to unravel the entire schedule precisely because that allowance exists; strip it out, and every deviation transmits directly to the promised delivery date. The buffer embedded in setup time is therefore rational to the extent that it lowers short-run cost in a high-variability environment. The difficulty resides not in the allowance itself but in its persistence after variability has fallen, and in the absence of anyone charged with reclaiming it.
The allowance is never reclaimed because the sanctions attached to it are asymmetric. Overrunning the standard is a visible event, entering the shift report and demanding an account, whereas materially beating the standard, once acknowledged, becomes the following period's threshold — meaning strong performance generates evidence that will subsequently be used against the party that produced it. The typical behavior observed under this configuration is a clustering of realized durations just beneath the standard, with the lower tail of the distribution going unreported altogether. The figure consequently operates as a one-way ratchet: it travels upward and does not come back. Over several years, the spread between a plant's defined setup standard and its technically attainable duration can approach an order of magnitude.
The institutional cost of that spread surfaces first in lot-size arithmetic. Setup time is a direct input to the economic lot size calculation, and as the duration grows the calculation recommends larger batches, which in turn pull both work-in-process and finished goods inventory upward. The balance-sheet trace of this tendency usually hides not in the headline inventory balance but in the aging profile of that inventory and in the length of the cash conversion cycle: the same revenue is being generated with working capital tied up for a longer interval. The apparent uneconomic character of small batches also propagates into the commercial function — minimum order quantities imposed on customers rise, low-volume but high-margin orders are either declined or become impossible to price coherently, and the plant locks itself, without deliberation, into a single customer segment.
The second cost is a capacity base that appears narrower than it is. When defined setup durations consume a substantial share of total available hours, plant reporting shows utilization approaching its ceiling, and when a growth requirement arrives, the only visible answer becomes an additional line or an additional machine. At that point the matter ceases to be the diagnosis of an operational problem and becomes a capital allocation decision; the source of the constraint is coordination and measurement, while the response offered is a fixed asset. The business case presented to the investment committee looks internally consistent, since the utilization rate it relies upon has been derived from the standards resident in the system — and those standards constitute the unexamined foundation of the case rather than one of its tested assumptions. Under such a configuration, the capital committed may exceed the magnitude of the underlying problem by several multiples.
The third cost becomes visible once the company enters a transaction process. Seeking to verify the capacity claim set out in the information memorandum, a buy-side operational diligence team will typically re-time setups on the floor and identify the gap between the system standard and observed performance. That finding enters through one of two doors: it either forces a downward revision of the capacity claim and the growth plan resting upon it, or it requires the working capital normalization to be reconstructed to the seller's disadvantage. The latter travels more directly into the closing price, since the normalized reference level for inventory is among the most heavily contested items in any completion adjustment mechanism. Where a volume-linked earn-out exists, the same gap gets priced a second time, now through the attainability of the targets.
What neutralizes this tendency is not individual discipline or heightened awareness but a deliberate institutional redesign of the measurement itself, and that design has four separable components. The first is a two-clock recording regime, under which the interval the machine is stopped and the work that genuinely requires the machine to be stopped are logged separately, so that waiting items exit the setup duration and become visible to their actual owner. The second is the assignment of externally performable preparation work to a named accountable party; so long as responsibility for having tooling, gauges, and first-article approval ready before the stop begins rests with the operator, the measurement will not improve. The third is converting the standard into a parameter with a designated owner and a scheduled revision cadence — where it is not written down who may revise it downward, on what evidence, and at what interval, the number will continue to climb. The fourth is decoupling the measurement window from performance evaluation, since the only way to observe the true distribution is to open a period in which reporting it costs no one anything.
BEIREK's intervention in situations of this kind begins not by standing up an improvement program but by disaggregating which number the decision chain actually rests upon. Reviewing a capacity expansion decision at a capital-intensive facility, the standards from which the utilization rate was derived are reconstructed from two-clock observation on the floor rather than from the system record; waiting items are separated out of machine downtime, and the gap between attainable and defined setup duration is expressed directly in terms of lot size, inventory level, and cash conversion cycle. The question reaching the investment committee thereby stops being whether an additional line should be acquired and becomes what volume the existing line carries under which conditions.
The durable mechanism installed alongside it is a decision-record discipline: every revision to a setup standard is logged together with its rationale, the measurement it relies upon, and the party requesting it, and that record is reviewed on a quarterly rather than an annual rhythm. Because the same record constitutes the documented basis of the capacity claim, it provides defensible ground in a subsequent diligence exercise or credit committee review; the question the company never asked itself ceases to be the first question the reviewing party asks. The measure of institutional maturity here is not that the duration is short, but that the method by which it was determined can be demonstrated independently of the founder and of any single shift supervisor.
A plant's real capacity is not the sum of the machines it owns but the product of an institutional agreement about how the time those machines stand idle is to be measured; where that agreement is unwritten, capacity is a quantity determined by habit rather than by management. The question it would be reasonable to ask before an investment decision is settled is not whether a new line is required, but when, and by whom, the number representing the existing line's capacity was last actually measured.