When a capacity investment approved six months earlier is finally commissioned on a line running a second shift, the station cycle time measured beforehand is demonstrably shorter afterward: the equipment is installed, operator training is complete, and the station, measured in isolation, reaches precisely the throughput written into the justification memorandum. The plant's monthly shipment figure, however, separates from its pre-investment level only marginally, and that separation is usually small enough to disappear inside ordinary seasonal variation. The one tangible change visible on the floor over the same period is that work-in-process inventory has moved a station downstream from where it previously accumulated. Operations describes this as the line settling into a new balance, while finance asks nothing yet, since it is not obvious which line item such a question would belong to in the first place.
The same pattern recurs well outside physical production. A service organization that automates proposal preparation in order to compress turnaround will watch proposals leave faster while the count of executed contracts holds flat, the binding constraint sitting not in drafting but in the review capacity of the legal function, which was never inside the investment scope. A software group that acquires tooling to accelerate development, leaving the weekly cadence of testing and release untouched, produces a larger queue of finished-but-unreleased work rather than more releases into production. What these cases share is that the improvement is real and locally verifiable, whereas the unit in which it was measured is not the unit that governs what the system actually delivers to a customer or to the income statement.
The pattern has a name, shifting bottleneck — the migration of the constraint to another point in the system as the original constraint is relieved — and its mechanics are as plain as the observation that a chain carries only what its weakest link carries. What makes it opaque inside institutions is that no single table ever sees the whole chain. The output of a line equals the capacity of its slowest station; once that station is accelerated, system output rises only to the capacity of the second-slowest station, and no further. The real effect of an improvement on the system, therefore, is not the capacity gain at the station that was improved but the distance between the first and second constraints. Where that distance is narrow, a substantial investment produces almost no incremental output at all, however faithfully it performed against its own specification.
The reason this mechanism stays invisible at the institutional level is not ignorance but the architecture of measurement itself. The investment case is typically prepared by the owner of a single station, and the only magnitude that owner measures, is accountable for, and can defend under challenge is the efficiency of that station; end-to-end flow time across the line, by contrast, falls within no individual manager's budget responsibility. That ownership gap between the local efficiency measure and the system throughput measure is what makes an improvement locally provable and systemically unprovable at the same time. Post-investment reporting accordingly presents station capacity rather than shipments almost every time, and this choice is not evasive but rational, being the most defensible number the existing measurement system is capable of producing.
Relocation of the constraint is, under certain conditions, an entirely functional outcome, and not every migration represents a loss. Working along a constraint chain, undertaken in the right sequence, is a legitimate strategy that lifts total system capacity in stages. The difficulty lies not in the migration itself but in its being unanticipated, and therefore in the return on the first investment having been modeled as though the constraint would never reappear anywhere else. To the extent the destination of the constraint is known in advance, the investment ceases to be a single decision and becomes a sequenced program, with cash flow planned against that sequence. To the extent it is not known, the company drifts into a series of expenditures each of which was justified on its own terms and none of which was ever approved in aggregate.
The balance sheet expression of that series appears first not in fixed assets but in working capital. When the bottleneck moves, the buffer inventory that had accumulated immediately upstream of it moves with it; the aggregate inventory balance looks unchanged, or marginally higher, while its composition shifts from raw material toward work-in-process, whose conversion period to cash is structurally longer than that of raw material. Inventory turnover consequently fails to improve in the post-investment period and frequently deteriorates somewhat, and because that deterioration is never associated with the capacity decision, it is attributed instead to soft demand or to procurement policy. Order lead time does not compress either, since the commitment given to customers was calibrated to the slowest point on the line and holds at the same level even after that point has changed address.
The second institutional cost emerges in the credibility of capital planning. A second capacity request arriving at an investment committee must be defended on ground where the output increase promised by the first request did not materialize, and that ground raises the approval threshold irrespective of the technical merit of the new proposal. The predictable result is that the correct second investment is deferred, or approved with its scope trimmed, so that the constraint chain is never broken and the company operates for several periods in an intermediate state possessing neither the old capacity nor the targeted one. The cost of that intermediate state is typically tracked in no line item of its own, dispersing instead across overtime, expedited freight, and the commercial concessions granted on the customer side to protect delivery commitments.
The third cost becomes visible once the company enters a transaction process. Across the diligence table, the gap between declared nameplate capacity and realized shipments over the trailing three years is first read by a buyer as a demand problem; when technical review deepens and the gap is traced to a constraint chain, the question migrates to heavier ground, because the growth plan underwriting the price then requires further investment, and that investment is charged to the buyer's side of the arithmetic. The outcome typically observed is not a direct reduction in the valuation multiple but the attachment of conditions precedent and earn-out thresholds to volume commitments, leaving the seller contractually exposed on the throughput of a line whose second constraint was never identified.
The mechanism that neutralizes this tendency is not individual vigilance or a more rigorous investment memorandum but a change in the unit of measurement. Once capacity decisions are anchored to end-to-end flow time rather than station efficiency, the systemic effect of an improvement becomes visible before the capital is committed rather than after. Three components carry that shift. The first is the construction of a constraint sequence for the entire line, setting down in advance where the bottleneck will move once the first constraint is relieved and which capacity ceiling it will meet there. The second is the presentation of the investment case as a sequenced program rather than a discrete decision, so that the capital requirement of the second and third steps is seen by the same committee at the same sitting. The third is that the performance indicator travels with the constraint, the headline reported metric relocating whenever the bottleneck does.
BEIREK operates these three components as a distinct recording discipline in capital-intensive facility and program investments. Before commitment, the constraint map of the line is drawn, and for each constraint the capacity ceiling, the volume of capacity released once that ceiling is reached, and the position at which buffer inventory will then form are all committed to writing; that record forms an annex to the investment decision, and committee approval is granted against the expected change in end-to-end flow time rather than against station capacity. After commissioning, a fixed review cadence is maintained, under which the first magnitude measured following an improvement is shipments and flow time rather than station efficiency, and the question of whether the constraint actually migrated to the location the map predicted is settled by that measurement rather than by narrative.
The operational consequence of this approach is that an investment decision carries different meaning by role. For the plant manager it means that the efficiency of a single station is no longer a self-sufficient measure of success and that identifying the next constraint on the line becomes part of the mandate. For the finance director it means placing a sequenced capital program into the cash flow rather than a single request with an open-ended sequel. For the investment committee it means that the question asked in return for the approved amount concerns delivery lead time and inventory composition rather than nameplate capacity. The same record produces a defensive position later, in a transaction context, since both the source of the gap between nameplate capacity and realized shipments and the program that closes it are already documented.
The true capacity of a line is not the capacity of its most expensive machine but the size of an interval that sits in no one's budget responsibility, namely the distance between two consecutive constraints. The question governing the return on capacity investment is therefore not which station will be improved, but whether it is known in advance where the system will come to rest once that station has been improved. Where the question goes unasked before commitment, its answer emerges later, in the inventory line and in the delivery lead time, long after the decision has been made and long after the reporting that justified it has been filed.