Demand and Capacity Out of Register: The Structural Cost an Average Conceals

In the monthly operations review, capacity utilization tends to arrive as a single figure, and because that figure is an annual average, it withholds the very information that matters most about it. Across the same twelve months a plant may have opened weekend shifts in three, shifted outbound volume to expedited freight in two, and run a line at partial load in four; by year end all of this converges into one mid-range percentage, and the board member reading the schedule sees a business operating within reason. The three months of compression and the four months of slack do not, however, offset one another in any economic sense: the first produces shift premiums, scrap, and commercial concessions granted to protect a delivery date, while the second is depreciation carried by volume that never arrived. Averaging fuses two dissimilar events into one statistic, narrowing the discussion to whether the average itself looks acceptable.

Within the same organization, the case for expanding capacity and the case for leaving it unchanged are typically built from opposite data, and the inconsistency rarely surfaces. A request for a second line, an additional warehouse, or one more machining cell is justified on peak-month order intake; the argument that the same investment is unnecessary rests on annual average utilization. Both parties cite accurate numbers, yet they speak from different time scales, and the decision is taken without that difference ever being named. The predictable outcome is an investment either oversized against the peak or undersized against the mean, with the error becoming visible only several periods later, by which point it presents itself as a variance attributable to pricing pressure, mix deterioration, or a supplier issue — anything, that is, other than the scale question that was never explicitly framed.

The pattern has a name — demand–capacity mismatch, the divergence between the level of demand and the structure of installed capacity — and the operative word is structure rather than level. Mismatch is not solely a matter of quantity: the timing of capacity, its flexibility across the product mix, its commissioning lead time, and the ease with which it can be reversed must each correspond to the way demand actually arrives. A facility holding theoretical capacity comfortably above annual demand can be functionally short of it because of the weekly distribution of orders or the volatility of the mix. The mismatch is therefore not, as is frequently assumed, a forecasting problem to be resolved through improved accuracy; even under a perfect forecast, capacity acquired in a form that does not match the form in which demand presents itself leaves the divergence exactly where it was.

At the core of the mechanism sit two mathematically dissimilar regimes entering the same decision. Capacity is a step variable: a shift, a line, a cell, or a storage bay either exists or does not, with no intermediate value available, and each step carries its own lead time — one that, on the equipment side, frequently exceeds a full budget cycle. Demand, by contrast, is continuous, it fluctuates, and it carries a mix dimension. Placed side by side, these two regimes generate behaviour that is counterintuitive on paper yet observable on any operating day: as utilization approaches the upper band, waiting time extends not linearly but at an accelerating rate. Beyond a certain threshold, high utilization consequently ceases to function as an efficiency indicator and becomes a delivery-risk indicator, while the same number continues to appear in the report under a favourable sign.

The organizational layer reinforces the mathematical one. Demand commitments originate in the commercial unit, where success is measured by order intake; capacity commitments are made along the operations and capital lines, where success is measured by the absence of idle fixed assets. Each side behaving rationally within its own measure produces a systematic institutional bias: the commercial side treats capacity as a costless option, while operations treats holding capacity below committed demand as prudence. It should be added that this tendency is entirely functional wherever demand volatility is low and the mix remains stable; planning to the average is, under those conditions, genuinely the least expensive shortcut available. The difficulty lies not in the shortcut itself but in its continuation after the volatility or mix regime has changed — a change that is rarely announced and usually recognised only in retrospect.

The financial expression of the mismatch seldom sits where it is looked for. On the surplus side, the cost accumulates in depreciation expense, in fixed asset turnover, and in fixed charge coverage; where volume does not reach the level required to carry the fixed burden of the installed structure, gross margin is structurally suppressed, and that suppression is then explained by price competition or mix deterioration, causes unrelated to the actual one. Where the capital structure is debt-financed, the effect advances a further layer, since a covenant written on fixed charge coverage converts a delay in volume ramp directly into a compliance question rather than an operating one. On the shortfall side, the cost accrues first in the calendar and only afterwards on the balance sheet: safety stock held to defend lead time lengthens the working capital cycle, depresses inventory turns, and slows cash conversion.

The income statement traces of the shortfall side are more dispersed and therefore more easily missed. Overtime premium dissolves into personnel expense, expedited freight into logistics, and the margin forfeited on volume pushed to subcontract manufacturing into cost of goods sold; because none of these is tracked as a line item in its own right, the capacity shortfall common to all of them remains absent from management reporting. Contracts carrying delivery undertakings weigh the picture further, since a liquidated damages clause moves the cost of the shortfall out of operations and straight into the commercial result. The most expensive item, however, enters no ledger at all: the quotation not submitted because it could not be served, the order declined, and the customer who, unable to secure allocation, begins qualifying a second supplier all sit entirely outside the accounting system.

In a sale, partnership, or financing process, the mismatch becomes a first-day diligence question: what capacity will carry the growth being presented, and when and at what cost will that capacity be acquired. Buy-side behaviour here is predictable — where the growth case requires a capacity step in the near term, the capital cost of that step is discounted into the valuation, and the multiple is set below that of a comparable business of similar volume carrying headroom. In the same review, a request to add overtime and expedition costs back to normalised earnings as non-recurring items is generally refused once the counterparty demonstrates a recurrence pattern spread across three years; a recurring mismatch does not, by definition, produce a one-time expense. Where the timing of the step remains uncertain, the structure is commonly rebuilt as an earn-out or a condition precedent, leaving part of the consideration with the seller.

What neutralises the tendency is not a better forecast but the manner in which the decision is recorded and disaggregated, and the intervention can be constructed across four separable components. The first is the separation of demand into three bands by degree of commitment rather than its presentation as a single figure: volume under contract, volume of high probability but uncontracted, and volume that remains speculative. The second is the mapping of each band onto a capacity layer of correspondingly different reversibility — contracted volume earning permanent capacity, probable volume earning recoverable layers such as added shifts and subcontract arrangements, and speculative volume earning option-like preparation only. The third is the writing of trigger thresholds for movement between layers before the decision moment arrives, so that the argument is not reopened once a threshold is met. The fourth is the calibration of the review rhythm to the commissioning lead time of the longest-lead capacity item rather than to the budget calendar.

For these components to function, the decisive detail is that the capacity decision be recorded at the moment of proposal rather than at the moment of approval. Where the record states which demand band was relied upon, which mix assumption was used, and at what threshold the decision would be considered falsified, the review conducted eighteen months later ceases to be an argument about responsibility and becomes an exercise in calibration. The same discipline renders reversibility visible as a cost item in its own right: the unit cost of subcontract manufacturing exceeds that of the internal line, but the difference is not an inefficiency — it is the price paid for flexibility against uncommitted volume, and once named that way the terms of the decision shift, since the comparison is no longer between two unit costs but between a fixed commitment and a purchased option.

BEIREK's intervention in capital-intensive projects is built along precisely this line: the capacity decision is treated not as a single investment approval but as a timing problem calibrated among procurement lead times, contractual undertakings, and financing conditions. What is constructed in practice is a decision record holding the mapping between demand band and capacity layer in one place, trigger thresholds written in advance for movement between layers, and a review rhythm built backwards from the longest-lead item. On the contract side, the gap between the liquidated damages cap and the throughput that can actually be carried is measured before signature; on the financing side, the behaviour of a covenant written on fixed charge coverage during the period preceding volume ramp is taken into account. The output of the work is not a forecasting model but a governance architecture stating in advance the conditions under which the decision turns in each direction.

The capacity structure of a business is the materialised form of that business's belief about future demand, which is why a mismatch reads less as a planning defect than as the physical trace of a belief left untested across several cycles. The question worth putting on the table is therefore not whether capacity is sufficient for demand, but whether it has been written down which band of demand the installed capacity answers to, and at what threshold that band would be treated as falsified. Absent such a record, the answer is reconstructed after the fact, and reconstruction after the fact reliably favours whichever explanation is least costly to the party offering it.