In a production planning meeting, a request from the commercial side for smaller and more frequent runs is usually met with a single sentence: the line cannot change over that often. The sentence closes the discussion, because it is delivered in the grammar of a technical constraint, and technical constraints are not negotiated across a conference table. Yet when the same meeting turns to how long a changeover actually takes, the figure that surfaces is rarely the output of a measurement; it comes instead from the recollection of a senior setter, or from a master data field in the ERP system that no one present can date. Who entered it, under what product transition, and under what staffing and tooling conditions, is seldom known. The parameter entered institutional memory once, and has circulated ever since with the standing of a fact.
The circulating life of that parameter is longer than the life of the decisions built upon it. Lot sizes are derived from it, minimum order quantities are derived from lot sizes, quoted lead times are derived from order quantities, and safety stock levels are derived from lead times. Every item at the far end of that chain occupies a visible line on the balance sheet, while the assumption at the head of it appears nowhere, being no accounting item at all. The predictable consequence is that annual budget reviews debate the inventory level at length while leaving untouched the mechanic that produces the inventory level.
The method that belongs here, named at this point rather than at the outset, is SMED — single-minute exchange of die, the compression of die and model changeover into the single-digit minute range. Its core distinction is simple, and draws its force from that simplicity: changeover work divides into internal setup, which necessarily occurs with the machine stopped, and external setup, which can be completed while the machine is still running. In many plants a substantial share of what is performed after the press comes down — retrieving the die from storage, staging fixtures, hunting for bolts, printing the work order — is technically external work performed internally. The first move toward a shorter changeover is therefore not new equipment but a reclassification of the work already being done.
The second layer of the method decomposes what remains genuinely internal, and the resulting distribution tends to sit uncomfortably with managerial intuition. Clamping and unclamping typically account for a comparatively modest share of the elapsed time, whereas adjustment and first-article approval — the trial cycles run until dimensions hold — account for the dominant share. Adjustment time is variable to the extent that it depends on the experience of the individual setter, and becomes fixed to the extent that it is transferred onto positioning references, stops, and standardized shims. Once that distinction is understood, changeover duration becomes visible for what it is: an output of method design rather than a property of the machine.
The behavior observed in most plants nonetheless treats changeover duration as given data and solves the lot-size question around it. Considered against the incentive structure beneath it, this behavior is coherent. Where production management is measured on output per shift, asset utilization, or OEE, every changeover is recorded as downtime, and the fastest route to less downtime is fewer and longer runs. Carrying cost, obsolescence exposure, and cash conversion appear nowhere in that same manager's performance definition; those items live on the finance side of the reporting pack. The issue, in other words, is not a wrong decision but a decision whose cost lands where the decision maker has no line of sight — a question of how authority and measurement are distributed rather than one of judgment or diligence.
The first institutional cost of this configuration accumulates in working capital. Long runs enlarge finished goods produced ahead of demand and the work in process held to keep the line fed; inventory turns slow, the cash conversion cycle lengthens, and during periods of growth the company experiences liquidity strain while remaining profitable on paper. That strain is commonly diagnosed as a financing problem and addressed on the financing side, through a larger revolving facility, extended supplier terms, or tighter collection pressure. A portion of the capital tied up, however, is the direct output of a lot-size parameter in the production plan, and is capable of being released within a few months if that parameter changes.
The second cost sits on the commercial side and accumulates more quietly. A high minimum order quantity systematically screens out low-volume, high-margin accounts; after a period, the sales organization stops calling on them as well, since an inquiry that cannot be converted into a quotation is time spent without return. The portfolio drifts, without anyone deciding that it should, toward a handful of high-volume and lower-margin accounts, and customer concentration rises. Concentration is itself a negotiating problem: in annual price discussions, a manufacturer drawing a material share of revenue from a single buyer holds a weaker position precisely because that buyer is aware of the share.
The third cost surfaces in the capital expenditure pipeline. When demand rises, the first reflex is a capacity investment — an additional line, a second press, an incremental shift. Yet if no one has measured how much calendar capacity is consumed by changeover, the decision is taken without knowing whether the bottleneck arises from processing speed or from setup load. Where changeover duration can be reduced by an order of magnitude, the effective capacity recovered on existing equipment may approach the increment an additional line would deliver, and does so without generating depreciation, an installation schedule, or incremental headcount. Whether that comparison appears in the file placed before the investment committee says a good deal, on its own, about the quality of the file.
These three costs converge into a single heading at the moment the company changes hands or becomes the subject of external financing. A capacity representation that holds only under a particular product mix and a particular lot size is a conditional capacity; where the condition goes unstated, it emerges during diligence, and at the moment of emergence it produces a cost in credibility that extends beyond the finding itself. Provisions taken against slow-moving inventory are, in the same way, the absence of flexibility as it has already been recognized in prior-period earnings. The typical consequence of such findings is not a headline price break; more frequently observed is a hardening of the working capital adjustment in the closing mechanics, a lengthened earn-out horizon, or a separately negotiated escrow tranche attached to the inventory line.
The mechanism that neutralizes this pattern is not greater care on the part of the operations team but the institutionalization of the data on which the decision rests, and it separates into four components. The first is definition: changeover is measured as the elapsed clock time between the last good part of the outgoing product and the first good part of the incoming one, with a single definition applied plant-wide, and with intermediate cleaning, breaks, and waits for quality release included rather than excluded. The second is classification: each changeover is logged with its internal and external components identified, so that improvement potential appears as a standing balance rather than as an aspiration. The third is authority: changes to lot size and minimum order quantity become a joint decision between production and commercial rather than a unilateral production preference. The fourth is cadence: these parameters are retested on a defined review cycle instead of resting permanently in a master data field.
BEIREK's intervention in configurations of this kind concentrates on establishing the decision base before any improvement program is run. We install a changeover log with defined start and stop moments, separate the internal and external content of that log by product transition rather than in aggregate, and render the resulting effective capacity table readable on the same page as the working capital position and the quoted lead time commitments. On capital expenditure requests, we define a threshold the file must clear before it reaches the investment committee: the justification for an additional line is not complete until the changeover load on existing equipment has been measured. The threshold is not a delaying device; it widens the set of alternatives the committee sees and makes explicit which assumption the recommendation rests on.
The flexibility of a manufacturing operation depends less on the vintage of its machine park than on which magnitudes are treated as open to question. From the moment changeover duration circulates as an unquestionable constant, every decision derived from it — run length, inventory, lead time, customer mix, capital expenditure — is taken in the shadow of that constant, and the company remains loyal for years to an operating model it never actually chose. The question worth asking, accordingly, is not how fast the line runs, but which parameter in the production plan was last measured, when, and by whom.
Where that answer is not known, every representation made about the plant's capacity is, in substance, a representation about a record held in memory.