In a quarterly planning meeting, the commercial side puts an incremental volume request from an existing customer on the table and asks operations for a single answer: can the volume be absorbed? Operations, looking at the annual nominal capacity figure it holds, gives a reasonable answer; the figure is real, it has been audited, and it appears in identical form in the investment file. That figure, however, was calculated on assumptions about a particular product mix, a particular shift pattern and a particular changeover frequency, and none of those three assumptions is retested in the room. No one misstates anything. The answer is simply given at the wrong resolution. The commitment is made not to the plant but to an average year of the plant, whereas the customer takes delivery not from an average year but from a specific week.
The second observation is that the condition rarely announces itself as a capacity symptom; it presents as a scheduling symptom. The expedited freight line grows quietly, weekend shifts stop being exceptions and settle into the routine, changeover counts climb, and small-volume orders drift backwards in the schedule without ever being formally declined. Monthly reporting still shows plant utilisation inside a comfortable band, because what is being averaged is the occupancy of every line across an entire month, while a single station runs at its effective ceiling during particular weeks of that month. The average is a statistic that conceals the constraint, and for as long as the constraint remains concealed, commercial commitments are made as though it did not exist.
The mechanism worth naming at this point is capacity shortage — the inability of production or logistics capacity to meet demand — although the great majority of cases encountered in practice are not deficits of total capacity but failures of constraint alignment. Capacity is not a scalar; it is a curve that is a function of at least four variables: product mix, changeover and setup time, equipment availability, and the number of operators holding the relevant certification. When the mix becomes heavier, the same line yields fewer units; when changeovers become more frequent, the net time available for production contracts; when a single critical station enters its maintenance window, the output of the entire line is set by that station. A capacity figure expressed without a stated validity window is therefore unusable for decision purposes, however accurate it may be in itself.
The second property of the curve is that its behaviour near the ceiling is not linear. Once utilisation rises above a certain band, waiting time increases disproportionately rather than proportionally, and the same plant invariably purchases the final few points of utilisation with lead time. Adding demand and process variability sharpens the picture further: of two plants carrying identical average demand, the one with higher variability will miss committed delivery dates with some regularity even though it appears adequate when assessed on averages alone. Capacity decisions are consequently governed by the tail behaviour of the queue rather than by its mean, and it is precisely this layer that planning discussions omit most often.
The tendency to systematically under-purchase capacity arises not as an error but as a rational preference that lowers cost under a specific set of conditions. Reserve capacity is directly visible in the expense statement — depreciation, rent, labour retained but not fully utilised — and that visibility exposes it to challenge in every budget cycle. The cost of capacity shortage, by contrast, carries no line bearing its own name; it disperses into expedited freight, overtime premium, liquidated damages, rework and orders that are lost without ever being recorded as losses. One side of the ledger therefore shows a single consolidated visible cost, the other an invisible cost distributed across five headings, and the decision predictably goes against the item that can be seen. The difficulty lies not in the preference itself but in the preference remaining fixed once the demand mix or the commitment horizon has changed.
The financial trace of this tendency is read not in the capacity line but, more often, in the composition of logistics and personnel expense. The share of expedited freight within total shipping cost is a more reliable indicator of the quality of constraint management at a plant than the utilisation ratio is; in the same way, the ratio of overtime premium to standard labour describes where the capacity curve actually terminates more accurately than any management representation. On the contractual side, exposure accumulates against the liquidated damages cap, and a delivery programme approaching that cap constitutes not merely a penalty risk but a lever the customer will hold in the next pricing negotiation. Customer concentration deepens over the same period, since priority under constraint is granted to the largest account while smaller customers fall out of the portfolio without any decision ever being taken to release them.
Even where a capacity gap remains invisible in a company's own reporting, it becomes legible at the diligence table during a sale or a financing process. When the buy side compares the delivery commitments in the order book against actual output over the trailing twelve months, a band in which commitment exceeds demonstrable capacity typically emerges, and that finding returns as a direct discount applied to the growth plan, a price adjustment, an increased escrow ratio, or an earn-out threshold anchored to a volume that cannot be reached. On the lending side the effect is harsher: where the DSCR calculation rests on a volume assumption above demonstrable output, the credit committee treats the matter not as a question of financial assumption but as an absence of operational evidence. What determines value here is not performance itself but the demonstrability of performance being repeatable at the committed volume.
This tendency is neutralised by institutional architecture rather than individual attention, and the architecture separates into four components. The first is that capacity is declared not as a single annual number but as a curve stated at the level of the constraining resource, conditioned on product mix and accompanied by a validity window. The second is that every commercial delivery commitment is written, at the moment it is given, into a commitment register that ties it to a bottleneck load check. The third is that the rhythm of the load-versus-capacity review is calibrated to the longest lead time for adding capacity — whichever of tooling lead time, operator certification curve, permitting cycle or equipment order runs longest sets the review horizon, not the monthly calendar. The fourth is that the utilisation band triggering escalation is defined in advance, so that the threshold fires when queue behaviour begins to deteriorate rather than after a delivery has been missed.
The same structure means different things to different roles, and where that distinction is not observed the mechanism remains a paper exercise. For the operations director the question is whether the constraining resource is measured at weekly resolution and whether the maintenance window is planned against that measurement. For the commercial director it is whether the delivery date has ceased to be a negotiating variable and become an output of the capacity model. For the finance director it is whether the cost of reserve capacity is compared, within a single statement, against the sum of expedited freight, overtime and penalty exposure, since as long as those figures live in separate statements the comparison is never actually made. For the lender it is whether a single document makes visible whether the debt service assumption rests on demonstrable output or on nominal capacity.
BEIREK's intervention in this problem begins by reconstituting capacity as a matter of commitment governance rather than production reporting. We build a capacity model at the level of the constraining resource, sensitive to mix and explicit as to its validity window; we operate a commitment register that binds commercial approval to that model; and we review the gap between load and capacity on a fixed rhythm calibrated to capacity lead time. The function of that register is not to audit the past but to make visible, ahead of the decision point, the month in which a capacity investment decision has to be taken. On the contract line, we align the delivery programme, the liquidated damages cap and the definitions of force majeure and excusable delay with demonstrable output; absent that alignment, the difference between what the commercial team signs and what the plant can actually produce converts quietly into penalty exposure.
Capacity shortage is seldom the absence of a machine; it is usually the absence of any record connecting a commitment to the constraint that must honour it. The real capacity of a plant is visible not in its nominal figure but in the date of the last meeting at which someone asked which mix and which week that figure was valid for.