In a monthly planning meeting, the items that went out of stock are opened one by one, and for each of them a plausible explanation surfaces — a supplier delay, an unanticipated spike in demand, a shipment held at customs, a shift in the production schedule. Having found the explanation, the meeting advances to the next item; the question of when the triggering threshold was set, on which assumption, and under whose authority is typically never raised, since the threshold is treated not as a decision but as a property of the system. Yet in many organizations that number was loaded during ERP implementation as a provisional parameter set, left uncalibrated in the pressure of go-live, and has since remained in place as a figure no one has claimed. To the extent that the threshold has no owner, the outcome it produces enters no one's performance review.
The side of the same meeting that is never opened is the surplus. How many months of coverage sit in the warehouse behind an item that has not gone out of stock for a year does not reach the agenda, precisely because no one generates a complaint about it; a stockout produces a phone call, an escalation, and a customer email, whereas idle inventory occupies a rack and stays quiet. This asymmetry in feedback pushes the parameter set in one predictable direction over time — upward — since every stockout event resolves into a raised threshold while no surplus event resolves into a lowered one. Viewed item by item, the resulting picture reads as a collection of disconnected decisions; viewed at the level of the enterprise, it has the character of a systematic drift.
The name of this pattern is **reorder-point failure** — the replenishment order releasing at the wrong level or at the wrong moment — and its mechanics rest on the fact that a threshold, while expressed as a single number, carries three separate estimates simultaneously. The reorder point depends on the lead time, on the demand expected to materialize within that lead time, and on the probability of stockout the organization is willing to accept for that item, which is to say the targeted service level. Although all three inputs are estimates, the output is read not as an estimate but as a threshold, and therefore as a fact; from the moment a number acquires the status of fact, it closes to debate. This is why a mistaken threshold is seldom discussed as a threshold, and almost always discussed as supplier performance or forecast accuracy.
The shortcut is not itself an error. In an organization carrying thousands of stock-keeping units, opening every order on individual judgment is not feasible, and threshold logic lowers the cost of decision by an order of magnitude, making the operation workable at all. The problem lies not in the shortcut but in the shortcut remaining fixed while the conditions underneath it move. As realized shipping durations widen seasonally, as minimum order quantities or container consolidation practices change, as demand variance in the product mix rises alongside a new customer segment, all three inputs to the calculation shift, while the threshold stands at the value first loaded. Under those conditions the threshold no longer represents a service level decision; it represents an assumption made years earlier and subsequently forgotten.
The second and less noticed mode of failure lies not in the level but in the timing. Where triggering logic operates on physical stock without netting open purchase orders and reserved quantities, the system either releases a duplicate order or delays one it should have released. Layered on top of this is the fact that review is periodic rather than continuous: if stock position is checked weekly, the effective threshold is not the documented one but the documented one plus a review period of demand, and that difference is written down nowhere. Batching purchase orders for freight economics produces a comparable displacement, in which the trigger forms at the correct moment while the order waits for the shipping window to open.
The balance sheet expression of this mechanic appears not in the aggregate inventory figure but in the distribution of days of coverage across items. Total inventory turnover may sit at an acceptable level while, inside the same portfolio, one group remains motionless for months and another is sustained by continuous expedited replenishment; the average conceals the tension between these poles. In working capital terms this means that the financing of a stock mass which does not convert to cash is being carried by the profitability of the portion that does, which in turn pushes the organization toward incremental financing at every step of growth.
The income statement trace is more dispersed, and it is that dispersion which keeps it from ever being aggregated. Costs arising from a late trigger scatter across four or five separate accounts as air freight premiums, partial shipment charges, expediting payments to suppliers, setup losses on the production line, and overtime; since none of them individually crosses a materiality threshold in management reporting, the sum is never seen. At the opposite pole, the customer-facing counterpart of a late trigger is a decline in on-time-in-full delivery, a metric tied in many corporate supply agreements directly to a penalty clause and, in some cases, to removal from an approved vendor list. The counterpart of early triggering surfaces as an inventory write-down provision, typically recognized with a lag of two or three years.
In a sale or investment process this layer inevitably comes to the surface at the diligence table. Where quality of earnings analysis isolates expedited freight buried inside cost of goods sold, the buyer tends to normalize those charges as recurring rather than one-off, on the reasoning that what produces the expense is a parameter regime rather than an event. The more expensive second effect emerges in the setting of the working capital target: a target built on historical averages absorbs the inflated inventory level, obliging the seller to deliver that level at closing, so that the cash tied up in excess stock is effectively transferred without consideration. The material loss in valuation is more often realized in that negotiation than in the multiple.
The mechanism that neutralizes this tendency is not individual vigilance but the restoration of the parameter to decision status, and it has four separable components. The first is ownership: every parameter group has a named owner, and no threshold change enters the system without a record of who made it and on what grounds. The second is measurement: the input to the calculation is not the lead time written into the contract but the distribution of realized duration from purchase order date to goods receipt, and the tail of that distribution is monitored alongside its mean. The third is segmentation: rather than applying a single default service level across the whole portfolio, the target is chosen explicitly by item group according to margin contribution, sourcing flexibility, and the customer-side consequence of a stockout. The fourth is cadence: review occurs at a frequency governed by the rate at which the lead time distribution moves, not by the calendar.
BEIREK's intervention in this area is not to assume individual item decisions but to rebuild the frame in which those decisions are produced. The first structure established is a parameter register: for each stock group, the threshold, the three assumptions that generated it, the measurement window those assumptions rest on, and the rationale for the most recent change are held in a single record, so that the threshold ceases to be a system property and becomes a traceable decision. On that foundation, an exception cadence is operated in which realized lead times and demand variance are re-measured periodically and items whose deviation exceeds a defined band fall automatically onto the review agenda; the agenda carries not the entire portfolio, but only what has drifted.
The second line of intervention sits in the authority architecture: the role that releases an order is separated from the role that can alter a threshold, and the setting of the service level target is lifted out of the operational tier and made a joint decision with the finance tier that bears the working capital consequence. That separation renders two different outcomes of the same decision — delivery performance and trapped cash — visible at one table, and structurally rebalances the feedback asymmetry, since the noise of a stockout and the silence of excess stock equalize only at a tier that carries the cost of both. Recording the rationale at the moment of proposal rather than at the moment of approval makes it possible, at the next review, to test whether the assumption still holds.
The maturity of an organization's inventory management is legible less in the frequency of its stockouts than in its capacity to show when and why its thresholds were changed, since the first is a function of external conditions and the second a function of the firm's own decision architecture. So long as the reorder threshold remains a system parameter, it is nobody's error — and for precisely that reason, it is corrected by nobody.