When the monthly service-level report reaches the table in an operations review, the figure on the sheet tends to read comfortably, while the account given by the commercial lead present in the same meeting, describing the same period, rarely confirms it; partial deliveries accumulating on the customer side, supplementary dispatch notes opened for a second consignment, split invoicing and repeated follow-up correspondence together produce an experience that does not match the performance the report describes. The distance between these two accounts seldom arises because one is correct and the other mistaken — each is accurate within the definition it uses. The divergence originates in the unit the ratio rests on: a percentage computed over quantities shipped runs systematically higher than one computed over orders delivered complete in a single consignment, and the spread between those two numbers widens as the portfolio broadens. Nobody in the room has said anything untrue; the same phenomenon has simply been observed from two different surfaces.
A second expression of the same pattern surfaces in order-acceptance behaviour as the period closes. Approaching month end, deferring entry of an order for a line with no inventory behind it, writing the delivery date into the following period, or steering a customer request toward a substitute item are, taken individually, defensible operational choices made by people acting in good faith. In aggregate, however, they improve the measured ratio while removing the unmet demand itself from the measurement surface. What produces this behaviour is not an individual intent to manipulate a number; a performance indicator defined on a period basis pushes the decision-maker predictably in that direction, and the record simply follows the boundary it has been given.
The mechanism that warrants naming at this point is **fill-rate failure** — the inability of the first shipment to satisfy the targeted proportion of demand, with the residual closed through partial delivery, backorder or substitution — though the mechanism itself amounts to considerably more than an inventory shortfall. Fill rate is, by construction, an *intersection* measure: it registers the degree to which the stock held at the moment demand arrives overlaps with the composition that demand requires. Total inventory may therefore be entirely adequate while the ratio still deteriorates, to the extent that the distribution of that inventory across line items diverges from the mix being requested; a company can carry excess stock and a weak service level simultaneously, and the coexistence of those two conditions is not a contradiction but a typical outcome.
That the measure can be computed at three distinct levels is the variable that governs how visible the mechanism becomes institutionally. The unit-based ratio, dividing quantity shipped by quantity requested, yields the most generous result in high-volume orders carrying many lines. The line-based ratio, weighting each line within an order equally, responds more sensitively to distortions in the inventory mix. The order-based ratio requires the entire order to ship complete and in a single movement; this is the measure the customer actually experiences, and it is ordinarily the lowest of the three. Unless a company fixes explicitly which of these it reports, the inconsistency between its claim of improvement and the complaint arriving from the customer becomes structural rather than episodic.
There are conditions under which the mechanism is genuinely functional, and failing to see them leads to framing the problem incorrectly. Partial shipment creates real value where the customer needs the critical line early rather than waiting for completeness; with a maintenance item, a spare holding a production line idle, or project material against a narrow window, waiting for the full order to assemble is the worse outcome for the receiving party. By the same logic, targeting complete delivery on every line in a slow-moving, high-variety portfolio inflates inventory cost out of proportion to the actual structure of demand. The difficulty lies not in the shortcut itself but in the shortcut persisting once the condition has changed: a partial-shipment practice designed for urgent items, once it becomes the default response for every plannable line, ceases to function as a flexibility instrument and becomes a shadow cast across the measurement.
On the balance sheet, the counterpart of this tendency is usually concealed not in the inventory line itself but in the composition of that line and in the length of the working capital cycle. An unmet first shipment generates a second pick, a second pack, a second freight movement and, in most cases, a second invoice; because these costs are not gathered under a single account, they disperse within logistics expense and cannot be traced back to their cause. At the same time, the safety stock held to compensate for the recurring shortfall depresses turnover, yet fails to improve service precisely to the extent that it accumulates against the wrong items. The company ends up carrying both the capital tied into inventory and the cost of repeated handling, and because each line item furnishes a justification for the other, neither is examined.
On the commercial contract surface the cost appears in sharper form. Framework agreements with institutional buyers increasingly write the service commitment against a defined threshold, and the remedy triggered below that threshold takes the shape less often of a direct penalty than of a price revision right, loss of sole-source status, or the buyer's right to onboard a second supplier. What these three share is that the cost registers not as an expense line in the income statement but as a structural narrowing of the following period's revenue. Where the high ratio appearing in the supplier's own reporting rests on a different definition from the ratio the customer computes in its own system, that divergence produces a bargaining asymmetry against the supplier at the negotiation table; to the extent the counterparty fixes the definition of the measure, the discussion has concluded before it begins.
The valuation counterpart of the same phenomenon typically emerges during an acquisition process or financing diligence under the heading of customer concentration. The reviewing party looks less at the revenue figure than at how dependent that figure is on contract renewals; the distribution of first-shipment performance by customer offers a more reliable indication of the renewal probability of the three largest accounts than the commercial team's own characterisation of the relationships. Where the recurring shortfall is seen to concentrate in specific customers, the response is usually not a direct multiple reduction but an earn-out structure tied to post-closing revenue continuity, or an expanded set of representations and warranties. The question the company has generally never put to itself is a narrow one: who calculates the measure, under which definition, in which system, and does that calculation reconcile with the customer's own record.
Structural intervention begins with anchoring the definition of the measure in a single place, since no improvement effort rests on verifiable ground while the definition continues to drift. This correction has four separable components: first, fixing the reported ratio to an order basis while retaining the unit-based ratio only as a secondary diagnostic; second, taking the stockout record not at the moment of dispatch but at the moment a delivery date is promised to the customer, so that demand refused or deferred also enters the measure; third, removing the partial-shipment decision from operator discretion and binding it to a defined item classification, meaning that the groups in which split delivery is acceptable are written down in advance; fourth, reallocating the safety-stock decision away from an aggregate value and toward item-level demand variability and lead-time variability.
BEIREK frames a correction of this kind not as an inventory optimisation exercise but as a decision-architecture exercise, because the origin of the problem lies more often in which table the decision is taken at, against which record, and at what cadence, than in the mathematics of the planning model. The first mechanism established in practice is a single reconciled measurement definition together with a commitment record attached to it: the record taken at the moment a date is promised to the customer also captures the inventory assumption that promise relied on, and period-end performance is assessed against that assumption rather than against a retrospective narrative. The second mechanism is comparison, within a single review cadence, of the indicators reported separately by procurement, planning and sales; three functions producing different numbers for the same period is not a data problem awaiting reconciliation but an authority-distribution problem awaiting diagnosis.
The second layer binds the supply and commercial contract lines to the same definition. In most companies the service commitment in customer framework agreements and the lead-time and delay remedies in supplier contracts are negotiated independently of one another; yet the gap between the output promised and the input secured is the durable source of the first-shipment shortfall. The review we establish sets the two contract sets against each other in a single table and separates three points: in which item groups the commitment given to the customer exceeds the assurance obtained from the supplier, what portion of that gap should be closed with safety stock, and what portion requires the contract language itself to be rewritten. Until this separation is made, the relationship between inventory growth and service level is reopened for debate every period and concluded in none of them.
Demand that the first shipment cannot meet is less an operational indicator than a way of reading the distance between a company's own commitments and its own supply capacity; unmeasured, that distance does not disappear but merely disperses into logistics expense, into the inventory line, and into the contract renewal rate. The single question determining whether a management team can rely on its own service-level report is not how high the ratio stands, but whether that ratio reconciles with the number the customer has calculated in its own records.