--- title: "The Bottleneck: When One Stage Sets the Pace of the System, Why Capacity Budgets Get Distributed Evenly" description: "Total system output is set by the narrowest stage; capital directed anywhere else raises work-in-process inventory and working capital without adding a single unit of delivered volume. An hour lost at the constraint is lost for the whole system and cannot be recovered, while an hour gained away from the constraint is an accounting improvement rather than a throughput improvement." url: https://www.beirek.com/en/blog/bottleneck-capacity-constraint-management canonical: https://www.beirek.com/en/blog/bottleneck-capacity-constraint-management published: 2026-01-28 modified: 2026-01-28 category: "Operations & Supply Chain" category_url: https://www.beirek.com/en/blog/category/operations-supply-chain language: en-US reading_time_minutes: 8 publisher: BEIREK LLC publisher_url: https://www.beirek.com license: "© BEIREK LLC — citation with attribution and link permitted" keywords: ["bottleneck","theory of constraints","capacity investment","working capital","throughput","due diligence"] topics: ["Operations management","Capacity planning","Working capital and inventory","Manufacturing due diligence"] alternate_language_url: https://www.beirek.com/tr/blog/bottleneck-capacity-constraint-management --- # The Bottleneck: When One Stage Sets the Pace of the System, Why Capacity Budgets Get Distributed Evenly > **In short:** Total system output is set by the narrowest stage; capital directed anywhere else raises work-in-process inventory and working capital without adding a single unit of delivered volume. An hour lost at the constraint is lost for the whole system and cannot be recovered, while an hour gained away from the constraint is an accounting improvement rather than a throughput improvement. *The output of a sequential system equals neither the sum nor the average of its stage capacities but the capacity of its narrowest stage. Distributing a capacity budget fairly across departments enlarges working capital without moving delivered volume, and a company that cannot locate its own constraint cannot price its own growth plan.* --- Capacity investment discussions tend to arrive at the table in a recognizable shape. Each unit documents the utilization of its own station, the number of shifts it is running and the queue of work standing in front of it; the unit reporting the highest utilization is treated as holding the strongest case, and approvals are distributed along that ranking. By the close of the meeting the aggregate spend has been held inside an acceptable band and every unit has secured the machine or the headcount attached to its share. A year later, monthly delivered volume sits close to where it stood before the money was committed. Nobody made an error along the way — each request was internally coherent, each justification was documented — and the output of the system nonetheless did not move. The pattern is not confined to a production floor. Design packages queued behind a single approval signature in an engineering organization, proposal files stacked in front of a single technical review desk in a services firm, a specific component dependent on one qualified supplier in a procurement chain: each configuration produces the same structure. Work accumulates at one point in the system; the stages upstream of that point generate reports demonstrating how quickly and efficiently they are running, while the stages downstream, waiting for material to arrive, drift toward off-plan work in order to protect their own utilization figures. The weekly management agenda fills progressively with expedited freight, overtime authorizations and customer delay notices, none of which is the actual subject of the agenda. The mechanism underneath this behavior is the arithmetic of a serial system. In a chain of sequential stages, total output equals neither the sum nor the average of stage capacities but the capacity of the narrowest stage — the bottleneck, the stage that governs the pace of everything else on its own. Two consequences follow directly, both of them counterintuitive: an hour lost at the constraint is lost for the entire system and cannot be recovered anywhere downstream, while an hour gained at a non-constraint stage contributes nothing to system output and improves only that stage's unit cost report. Full utilization of a non-constraint stage produces, by definition, not throughput but work-in-process waiting in front of the constraint. The persistence of this mechanism owes less to inattention among decision makers than to the architecture of measurement itself. Utilization is inexpensive to measure, maps cleanly onto departmental accountability, and in a cost accounting convention that absorbs overhead across units produced, rising utilization makes unit cost appear to fall; maximizing local utilization is therefore entirely rational from the standpoint of the metric each stage is judged against. Under conditions where demand sits well below installed capacity and the order book is loose, that shortcut genuinely lowers cost and does no harm. The difficulty lies not in the shortcut but in the measurement architecture remaining fixed once conditions change — once demand approaches capacity and once delivery dates become contractual obligations rather than internal targets. Two misconceptions about where the constraint sits routinely delay diagnosis. The first is the assumption that a bottleneck is a physical object; more often than not the stage governing system pace is a rule rather than a machine — a batch sizing policy, an approval threshold routing every purchase above a given amount to a single committee, or a quality control step staffed on one shift only. Removing constraints of this kind typically requires no capital expenditure whatsoever, only a redistribution of decision rights. The second is the assumption that the constraint is stationary; once a constraint is relieved, the pace of the system migrates to the next narrowest stage, and organizations that never formally declare the new constraint continue planning for months against a limit that has ceased to bind. The balance sheet expression of this tendency usually surfaces not in the income statement but in the working capital lines. As non-constraint stages keep running at full tilt, the accumulation in front of the bottleneck converts into inventory; turnover slows, the cash conversion cycle lengthens, and during growth phases the company generates a continuous financing requirement while still presenting as profitable. Once expedited freight premiums, unplanned overtime, changeover time lost to reprioritization and the commercial concessions extracted after customer delays are added to that figure, the annual cost of leaving the constraint unmanaged will typically exceed, by several multiples, the capital cost of relieving it. Because these amounts are booked across scattered accounts, they never appear on a single expense line. On the commercial side the cost is paid in delivery reliability. When the sales organization commits to a schedule calculated from nominal plant capacity, the date written into the contract reflects the optimistic figure produced by summing stage ratings rather than the true pace of the system; liquidated damages, guarantee calls and the customer's own replanning costs are the direct consequence of that gap. The same structure deepens customer concentration wherever one large account's orders occupy the constraint continuously, since the allocation of constraint time is in substance a portfolio decision about which customers the company serves — and that decision is generally taken at the planning desk rather than at the commercial strategy table. At the diligence table the subject opens with a question about how the capacity claim was constructed. Where the information memorandum presents capacity as the sum of nominal stage ratings or as the highest historical monthly production figure, the question the buy side ought to put is narrower: at that output level, which stage reaches saturation, and to reach the volume assumed in the growth plan, which stage requires investment, in what amount, and on what lead time. The absence of a ready answer on the sell side typically affects price negotiation from two directions — the capital expenditure implied by the growth plan has not been embedded in the valuation, and the volume threshold attached to any earn-out may have been calibrated at a level the existing system cannot physically produce. Where the constraint resides in a single person — one competent operator, one process engineer, one authorized signatory — the same finding is written instead under founder dependency and moves directly into the scope of representations and warranties and the sizing of escrow. The mechanism that neutralizes this tendency is decision architecture rather than individual awareness, and it separates into four components. The first is locating the constraint by measurement rather than by opinion; the correct indicator is not utilization but the quantity of work accumulating in front of each stage and the time that work spends waiting, since a constraint reveals itself through the persistence of its queue rather than through how busy it looks. The second is subordination: the operating tempo of every other stage is set to the tempo of the constraint, and planned idleness at non-constraint stages is recorded as a design decision rather than as waste. The third is holding a protective time buffer immediately ahead of the constraint, a buffer deliberately concentrated at one point rather than dispersed as inventory across the line. The fourth is re-declaring the constraint after it has been relieved, which is a recurring rhythm tied to the budget cycle rather than a one-time analysis. None of these components survives unless it is recorded who declared the constraint and on what date. A single question placed in front of every capacity approval will often do more work than an entire analytical framework: does this investment relieve the constraint as it stands today, or does it enlarge the capacity of a stage that is not the constraint. Where the second answer is given, the investment need not be refused — some spending is defensible on grounds of resilience, redundancy or quality — but the justification should not be written as a throughput increase, because writing it that way builds the following period's targets on a capacity assumption that will not materialize. BEIREK's intervention in capital-intensive projects and multi-asset industrial portfolios concentrates precisely here. At project or facility level a single constraint register is maintained: which stage governs system pace, by what measurement it was identified, on what date it was declared and under what condition it will be revisited, all living in one document, and that document is a mandatory annex to the capital expenditure approval file. Across engineering and procurement lines the same logic is tied to the contractual schedule — long lead equipment, permitting and interconnection steps and individual signatures in the approval chain are listed as constraint candidates, and schedule float is concentrated ahead of the identified constraint rather than distributed evenly across every line item. On the operating side the re-declaration of the constraint is embedded in the monthly management rhythm, with the opening item of the meeting being not last month's output but which stage is setting the pace of the system this month. Once that rhythm is established, which stage the expedited freight and overtime authorizations are actually feeding becomes visible without further analysis, and that visibility frequently produces the first increase in output obtained without any additional capital. In transaction settings the same register allows the buy side to verify a capacity claim against the binding constraint rather than against a nominal total, which moves the price negotiation out of a debate about assumptions and into a debate about an investment schedule. The capacity of a manufacturing or service system is ultimately written not in the list of assets owned but in the narrowest link of that list. A company unable to name its own constraint, with a date attached, has not thereby eliminated the constraint; it has only established that the party deciding where the constraint sits is the system rather than the company. ## Key Points - When capacity investment is approved on the basis of departmental utilization rates, the funding predictably lands on a non-constraint stage and total output remains where it was before the spend. - One hundred percent utilization outside the constraint produces inventory rather than throughput, and that inventory appears on the balance sheet as working capital, lengthening the cash conversion cycle. - A bottleneck is not always a machine; a single qualified operator, a single approval signature, a single qualified supplier or one testing step can just as easily govern the pace of the entire system. - Constraints migrate, and an organization that does not formally re-declare the constraint after relieving the previous one will keep planning for months against a limit that no longer binds. - When a capacity claim is presented at the diligence table as the nominal sum of stage ratings, the capital expenditure implied by the growth plan has almost certainly not been priced into the offer. ## Questions ### How is a bottleneck identified in a system? The correct indicator is not stage utilization but the quantity of work accumulating ahead of each stage and whether that accumulation persists. A constraint reveals itself through a queue that does not systematically clear rather than through a high utilization figure. Utilization measurement misleads, because a non-constraint stage can also run at full occupancy by overproducing; what it produces in that case is not throughput but inventory waiting in front of the constraint. ### Why has output not increased despite our capacity investment? The investment was in all likelihood directed to a stage that is not the constraint. Because the output of a sequential system equals the capacity of its narrowest stage, capacity added elsewhere does not change the pace of the whole; it only enlarges the work-in-process accumulating ahead of the bottleneck and, with it, the working capital requirement. Decision processes that distribute the budget across units according to utilization produce this outcome predictably. ### Is idle time at a non-bottleneck station considered waste? No; planned idleness at non-constraint stages is a design decision. Running those stages faster than the constraint does not raise output, it creates inventory and lengthens the cash conversion cycle. Treating that idleness as waste is a measurement artifact of cost accounting that absorbs overhead across units produced. Sound practice sets the tempo of non-constraint stages to the tempo of the constraint and defines the unused capacity explicitly in reporting. ### How should a company's capacity claim be verified during diligence? Neither the sum of nominal stage ratings nor the highest historical monthly production figure is a binding indicator. Verification proceeds through which stage reaches saturation at the target volume, what relieving that stage costs, and what lead time it carries. Where the sell side has no ready answer, the capital expenditure required by the growth plan has not been reflected in price, and the attainability of any earn-out threshold is directly affected. --- Source: https://www.beirek.com/en/blog/bottleneck-capacity-constraint-management Publisher: BEIREK LLC — https://www.beirek.com