In a supplier review meeting, most lines on the approved vendor list carry two or three names; read further down, however, a defined subset of critical inputs shows a second name against which no purchase order has been placed in several years. That second name sits in the file, counts as an alternative in the audit record, generates an entry in the quality system, and has never once run on the production line. In the same meeting the incumbent's performance indicators are generally favourable — delivery discipline is high, scrap rates are low, the engineering teams on both sides know one another — and those indicators function as a silent justification for why the second name was never tried. The single figure absent from the pack is how many months would elapse before that second name could reach serial-production readiness were the incumbent to become unavailable; more often than not, the number has never been calculated.
A second pattern shows up in how the purchasing function is measured. Where purchasing performance is assessed principally through unit price, payment terms and continuity of supply, consolidating volume with a single supplier improves all three indicators simultaneously, whereas keeping a second source alive degrades each of them to some degree. The dependency therefore reports as a performance gain, and the effort to reduce it reports as a performance loss. What emerges is a structure that nobody inside the organisation is required to defend and that the measurement system actively rewards. So long as the input keeps flowing without interruption, that structure discloses its cost in no table anywhere in the business.
The mechanism at work here is known in supply-chain practice as single-source dependency — the concentration of a critical input on one supplier — and it has a near twin from which it must be carefully separated. Sole-source describes the case in which no other producer exists on technical or legal grounds; patent protection, a singular manufacturing capability, or a restriction imposed by the customer's own approved list means there is no alternative to qualify. Single-source describes the case in which alternatives do exist but, never having been qualified, are unavailable in practice. The first is a market fact and is generally resolved only through design change; the second is an institutional preference and is manageable in full. In practice the two are frequently reported under one heading, and that consolidation causes the manageable case to be treated as though it were the unmanageable one.
The drift toward a single source is, under the conditions in which it arises, largely rational. Qualifying a new supplier requires samples, testing, field validation and, in many cases, customer approval; tooling, fixtures and test rigs must be paid for a second time; the shared technical vocabulary accumulated between engineering teams over years has to be rebuilt from the beginning. Consolidating volume, by contrast, improves the price tier, narrows the quality distribution, and accelerates joint development work. The difficulty therefore lies not in the shortcut itself but in the shortcut persisting after the conditions that produced it have changed: once product volumes scale, once the criticality classification of the input rises, or once concentration increases within the supplier's own customer portfolio, the same choice no longer generates the same economics.
A second layer of dependency forms behind the tier-one suppliers. An arrangement running two distinct vendors remains single-sourced at tier two if both draw on the same sub-component, the same raw-material processor, or the same specialised coating facility; diversity at the first tier renders singularity at the second invisible. That invisibility follows from the fact that supplier questionnaires are typically bounded at tier one. When a disruption does occur, both suppliers are observed to extend lead times at the same moment and for the same stated reason, and this simultaneity is interpreted internally as coincidence for a considerable period.
The first institutional cost of the structure accumulates in bargaining asymmetry. To the extent the counterparty recognises the switching cost and the qualification interval — and that recognition rarely comes late, since it is legible from the distribution of the order history — it will position itself accordingly in price negotiations, in requests to extend payment terms, and in capacity allocation decisions. This is not bad faith; it is the predictable output of the structure. The effect arrives not as a single shock increase but as small increments compounding across contract renewals; gross margin erodes modestly over several periods and the erosion is commonly attributed to a raw-material index or a currency movement. Unless the margin trajectory of single-sourced lines is tracked separately from that of multi-sourced lines, the divergence never enters the reporting at all.
The second cost appears in working capital. Once the dependency is recognised, the first reflex is typically to raise safety stock; that reflex reduces disruption risk in the near term while depressing inventory turns, lengthening the cash conversion cycle, and increasing storage and obsolescence costs. The result is a supply risk carried onto the balance sheet as an inventory line — the risk has not been eliminated, merely reclassified into a financing cost. The real information carried by the buffer is that however many weeks of production it covers, that figure is also the organisation's implicit estimate of how long a second source would take to activate; set the two numbers side by side and the buffer is usually found to fall materially short of the actual qualification interval.
The third cost surfaces on the diligence table. In an acquisition, a minority investment or a project financing, supplier concentration is typically treated with the same seriousness as customer concentration and it affects structure before it affects price: a separate heading opens in the representations and warranties, the escrow ratio increases, a second-source qualification plan enters the conditions precedent, and in certain transactions a portion of the consideration migrates into an earn-out tied to continuity of supply. The finding encountered most often at this stage is a change-of-control clause in the critical supply agreement requiring counterparty consent for assignment; the closing timetable of the entire transaction can come to rest on that single provision. On the project side the same dependency feeds directly into the construction programme, the liquidated-damages cap, and the scope of delay-in-start-up coverage in the insurance package.
What neutralises this tendency is not individual vigilance but a four-component institutional architecture. The first component is a dependency register in which critical inputs are consolidated in one record and every line is measured by a single metric — not the supplier's financial indicators, but the number of months required to qualify a second source and bring it into serial production. The second is keeping that second source alive on the line rather than on the list, through orders placed at regular intervals even where they appear economically pointless. The third is the contractual toolkit: buyer ownership of tooling and fixtures, escrow of technical data and process documentation, capacity reservation and allocation-priority provisions, and assignment rights on a change of control. The fourth is decision rhythm: the dependency register is reviewed not during procurement cycles but at design freeze and at the moment of bidding.
The rationale for the rhythm component is that single-source dependency is usually created at the engineering desk rather than the purchasing desk. The moment a specification names a single manufacturer's part number, a proprietary material class, or a process performable at only one facility, the sourcing decision has in substance been taken; when the purchasing function inherits it months later, all that remains within its reach is price negotiation. The effective point of intervention is therefore specification writing rather than supplier selection, and what is required at that point is modest — a marking against each critical specification indicating that a second manufacturer could meet it, or, where no such marking can be made, a record that this was decided as a matter of design.
BEIREK's intervention in this area begins with mapping critical inputs before FID on the capital-intensive projects it manages and fixing, for each critical line, the resupply interval as the sole governing metric; that interval is then compared against the total float of the corresponding activity in the programme, and the gap between them is budgeted as an explicit risk item. The same register functions as a checklist in the negotiation of supply agreements: tooling and fixture ownership, technical data escrow, capacity reservation and assignment provisions are put on the table as standard on lines carrying high dependency, and deliberately left aside on low-dependency lines so as not to inflate negotiating cost where it earns nothing.
On the operating side the register is kept live and reopened at defined thresholds — a design change, a step increase in volume, a change in the supplier's ownership, or a concentration signal detected at tier two. The question asked in that review is not whether the supplier is reliable but whether the relationship still produces the same economics under present conditions, because single-source dependency is, at the moment it is established, most often the correct decision, and it begins generating its cost only once the condition that produced it has disappeared. The maturity of a supply structure is measured not by how many suppliers it works with, but by when, and on what stated grounds, the decision to work with one was last taken again.