When a monthly production review reports the line's equipment effectiveness score at slightly above eighty percent, the discussion around the table almost never turns to how that figure was constructed, settling instead on how many points it moved against the prior month. Later in the same meeting, the commercial side explains that incremental volume cannot be served from the existing line and that a second line should therefore be brought forward for consideration. The two statements are made minutes apart, yet the tension between them typically goes unremarked: if the line is already surrendering a fifth of its theoretical capacity, a meaningful share of the incremental capacity being requested is already standing on the floor. The recurring pattern is that the capacity decision is taken not against the measured effectiveness score but against the shift supervisor's intuitive sense of how full the line feels.
A second pattern repeats itself in the same meeting. When the loss categories are presented, availability losses — breakdowns, changeovers, waiting on material — are itemised in detail, while performance loss appears as a single line carrying a modest percentage. Quality loss, in turn, is reported below the actual scrap rate, because a portion of the material separated at the end of the line is reworked and re-enters the count. The relative magnitude of these three categories reflects not the physical reality on the floor but the ease with which each of them lends itself to being recorded.
The mechanism at work here is OEE loss — the multiplicative erosion of equipment effectiveness through availability, performance, and quality shortfalls — yet what governs the institutional outcome is not the loss itself but the reference against which the loss is computed. Each of the three components carries a denominator, and those denominators are typically set by the very unit being measured: planned production time by the team that builds the shift schedule, standard cycle time by the engineering group that runs the line, and the quality threshold by the operations function that determines where rework stops being acceptable. Where calibration of the measure and performance against the measure sit in the same hands, the score can rise without the line running any faster; a loosened reference is sufficient.
This configuration is functional under certain conditions and should not be read as a management failing. Having standard cycle time set by the team that knows the equipment is the only workable route to measurement at all when the product mix changes frequently and defining a separate theoretical rate for every variant is impractical; defining a recording threshold for micro-stops is likewise rational, given that an operator required to log every stoppage measured in seconds would be unable to perform the work itself. The difficulty lies not in the shortcut but in its persistence after the conditions change: once the mix simplifies, once the line acquires automated data capture, or once the company reaches an irreversible threshold such as a capacity investment, a reference that was serviceable for a period becomes a source of noise that corrupts the decision.
The quietest component of the loss typically sits on the performance side, since the threshold defined for micro-stops — a limit of a few minutes in most plants — presents itself as a recording convention while functioning as the single strongest determinant of how much of the loss ever becomes visible. Stoppages falling below the threshold enter no loss category, leaving the divergence between actual and theoretical output unexplained, and an unexplained divergence is, over time, closed by revising standard cycle time upward. That correction does not eliminate the loss; it merely rewrites the definition so that the loss falls inside it. A comparable rewriting occurs on the availability side when planned downtime is removed from the denominator, with calendar-based utilisation dropping out of the report entirely.
The balance sheet consequence tends to appear not in the cost of production but in the following period's increase in fixed assets. Where a second line is approved without being tested against the demonstrated maximum output of the existing one, the company has financed a bottleneck that does not exist, raised its fixed cost base permanently, and made the depreciation charge contingent on the timing with which demand actually materialises. Because the reference used in modelling the return is the reported effectiveness of the existing line, the model inherits the same calibration problem. The business case presented to the investment committee is internally consistent while resting on an assumption that has never been tested outside the unit that produced it.
A second cost accumulates in working capital. Where output variability is high and that variability does not surface within the measured loss, delivery reliability can be defended only through inventory, and work-in-process and finished goods begin to function as a buffer absorbing the line's undeclared variance. The resulting deterioration in inventory turns is read on the finance side as a supply chain problem, when its origin is the micro-stop profile of the line. The same mechanism reaches the customer: adherence to delivery dates is purchased with a safety margin proportionate to the unpredictability of the loss profile, and the lead time quoted remains conservative precisely where it ought to be competitive.
The third and most expensive cost surfaces at the moment the company changes hands or approaches external finance. The question posed in buy-side or lender technical review is not the reported effectiveness score but the number of units the line actually produced over the trailing twelve months, and that single comparison exposes the distance between nameplate capacity and sustained output in one step. Where the gap reaches a material magnitude, the route typically taken is not a direct reduction in price but the attachment of the capacity claim to an earn-out or a post-closing performance condition, leaving the seller carrying the burden of proving the capacity of its own plant. What governs valuation at that point is not performance itself, but whether performance can be shown to be repeatable independently of the founder and of the shift team's intuition.
The mechanism that neutralises this tendency is built by changing the architecture of the measure rather than by sharpening operator attention, and it separates into four components. The first is that standard cycle time be set outside the unit being measured — anchored to the equipment manufacturer's nominal data and to the line's observed best quarter-hour of running performance — and be revisable only against a written justification. The second is that the micro-stop threshold migrate from manual logging to automated signal capture, so that the threshold ceases to function as a constraint of recording. The third is that calendar-based utilisation be reported alongside equipment effectiveness, rendering planned downtime a negotiable item rather than an invisible one. The fourth is that loss review be detached from the rhythm of the shift meeting and attached to the rhythm of capital allocation, so that no capacity request reaches the committee without its loss register.
BEIREK operates this intervention not by installing a measurement system in the plant but by converting the link between the capacity claim and its evidentiary chain into an institutional condition of approval. In preparing an investment request, we assemble the existing line's demonstrated maximum output, its loss categories, and the reference against which each of those categories was computed into a single register, and we use that register to separate how much of the requested increment can be met through recovery of existing loss from how much genuinely requires new plant. Every revision made to standard cycle time is recorded with its date, its justification, and the authority that approved it, since that record is the only document that supports both the auditability of the internal decision and the defensibility of the measure in any due diligence that follows.
The second line of intervention is the distribution of ownership over the measure. Keeping the roles that set the reference, take the reading, and are assessed against the reading out of a single reporting line — cycle time and threshold definitions residing with engineering, recording and verification with quality, and the translation of loss into a capital decision with finance — produces three signatures reading the same figure against different interests. Once that separation is established, equipment effectiveness ceases to be a performance indicator and becomes a capacity inventory, and the inventory stands behind the investment decision rather than in front of it. The review rhythm we apply follows accordingly: the loss profile is opened not monthly, but at the frequency of the capital allocation cycle and before the same committee.
In most companies, the answer to how much a plant can produce lies not at the physical limit of the equipment but in the institutional consensus about how that limit was defined; and for as long as that consensus remains unwritten, a company learns its own capacity only when it attempts to sell it.
The operative question is not how many points of effectiveness the line is running at, but who set the references that produced that figure and when those references were last tested from outside.