On an assembly line, or in a site warehouse when delivery of a listed item slips, the decision is rarely made at a procurement desk or in an engineering office; it is made at the point where the material physically runs out. In a conversation lasting a few minutes between a stores supervisor and a field engineer, an alternative already on hand and believed to serve the same function is put into service, the line keeps moving, the schedule holds, and the day closes. Nothing of that decision survives on paper, because in the mind of the person deciding, what occurred was not a change but a continuation — production kept running. The recurring pattern is consistent: decisions of this kind concentrate markedly in the final third of the programme and among items whose unit price falls beneath the approval threshold.
The same pattern surfaces in manufacturing through the grade of a fastener, the source of a coating, the batch of a resin, the conductor cross-section of a cable, or the composition of a gasket; on site it appears in a valve class, an insulation thickness, an anchor type. What these instances share is that the substituted item is seldom the visible primary equipment and almost always the secondary component holding that equipment in service. When the brand of a major package changes, everyone notices and the matter routes itself into an approval path without prompting; when the origin of a sealing element changes, the number of people who register it falls to one, and that person is the one who made the call. Records are not kept because the event is not judged worth recording, not because anything is being concealed.
The behaviour has a name — unauthorized substitution, the introduction into production or fabrication of a material or part absent from the approved configuration, without engineering review and without entering the documentation chain. The mechanism beneath it is not carelessness but a reasonably coherent line of reasoning: the decision maker compares the item in hand against the function it must perform, rather than against the document that defines it. Functional-equivalence intuition produces correct outcomes across most of engineering practice, since a bolt genuinely carries the load and a gasket genuinely holds the seal. The difficulty lies not in the intuition but in its scope, in that a functional test covers the physical behaviour of the component while leaving untouched the contractual, insurance, and certification commitments in which that component is embedded.
A second layer concerns the way the decision validates itself. A successful substitution emits no signal whatsoever: the line runs, the test passes, the delivery is made, and the absence of failure is read as evidence that the judgement was sound. The first substitution thereby lowers the threshold for the second, the second for the third, and after a handful of cycles what exists is no longer an isolated exception but an established practice. The variable governing the speed of that drift is not culture but the response time of the approval channel, since where the field's decision window measures two hours and the typical turnaround on a deviation request measures two weeks, the channel is functionally unavailable and the decision flows, predictably, into an undocumented route.
A third layer originates in how approval thresholds are constructed. Most procurement and change-management architectures tier approval authority by unit price, notwithstanding that no dependable relationship exists between the price of a component and the risk it carries. A gasket holding a pressure boundary may prove more critical than a housing costing a hundred times as much; the fill material in a fire barrier may govern the insurance position more decisively than the whole of the equipment in the room it protects. To the extent that a price-based threshold structure places precisely those low-cost, risk-concentrated items outside the approval perimeter, it makes undocumented substitution structurally easier rather than harder.
The first surface on which the institutional cost appears is warranty and insurance. Equipment warranties are typically extended against the approved materials list and the manufacturer's installation conditions, so the entry of an off-list component into the system tends to open coverage to dispute not for that single item alone, but across the subsystem the item affects. On the insurance side the consequence is sharper: the first document an underwriter requests in a claim file is the record demonstrating that the as-installed configuration matches the approved one, and an inability to demonstrate that match turns the subrogation relationship against the insured. The certification dimension is usually the last to be noticed, given that a listed assembly is listed as a specific component configuration rather than as a function.
In financed projects the cost becomes visible through the calendar. What the independent engineer or the lender's technical adviser does at milestone certification is, in substance, to compare the as-built documentation against the approved submittal set, and where an unexplained divergence between the two emerges, the outcome is ordinarily not rejection of the project but suspension of the certificate. A suspended certificate moves progress payment, release of the performance security, and in certain structures the COD date, all together; once the delay enters the LD calculation, the financial consequence generated by a component regarded as trivially priced reaches an order of magnitude that bears no comparison to the cost of the component itself. The source of that displacement is a documentary gap rather than a technical deficiency.
At the M&A table the same gap is priced in a different vocabulary. When a buyer's technical due diligence team samples production records, what it looks for is not an immaculate manufacturing history but the ability to establish which component, from which batch, entered which unit. Where that traceability exists, past substitutions are treated as manageable items and typically resolve into nothing more than a narrowed scope in the representations and warranties. Where it does not, the buyer prices not an identified defect but an unbounded liability tail, and the customary response is an increased escrow percentage, recall and field-remediation exposure tied to an earn-out trigger, and a records reconstruction exercise added to the conditions precedent.
The logic of the valuation discount clarifies at that point: the discount attaches not to the substituted part but to the company's inability to show where its own liability ends. A manufacturer able to confine a potentially affected batch to three hundred units holds the cost of any field intervention as a calculable line item; a manufacturer unable to confine it must calculate against the entire installed base. The difference between the two arises from two different standards of record-keeping rather than two different standards of engineering, and that is precisely the distinction buyers and lenders are drawing. Unauthorized substitution is therefore far less a quality question than it appears and far more an indicator of institutional maturity.
The mechanism that neutralises this tendency comes from system design rather than individual discipline, and it separates into four components. The first is that the record is opened at the moment of proposal rather than at the moment of approval: the field logs the contemplated substitution on a single screen before applying it, and that entry functions as the traceability record rather than as a precondition of approval. The second is the migration of the approval threshold from unit price to risk class, such that in categories including pressure boundary, fire barrier, grounding, and structural connection the threshold is zero irrespective of cost. The third is binding engineering review to a response commitment matched to field tempo — measured in hours for critical categories and in days elsewhere. The fourth is that every approved substitution carries an expiry, since an approval granted without a term quietly becomes permanent at the next design revision.
The intervention BEIREK builds along this line is not the drafting of a quality procedure but the placement of a recording surface at the point where the decision is actually made. On the projects we run, the substitution log sits within field execution rather than within the procurement system; each entry opens with references to the approved submittal set, the qualified supplier list, and the applicable certification scope, and no milestone is submitted for certification until the entry has been closed into the as-built set. The accompanying rhythm is a weekly reconciliation session in which the substitutions opened that week are compared across three columns — engineering disposition, supplier status, closure status — with the count of open entries reported to the sponsor as a project health indicator. The same structure carries different meaning for different parties: for the sponsor, predictability in payment certification; for the EPC contractor, preservation of warranty coverage; for the lender, a reduced volume of independent engineer findings; and for the eventual buyer, a traceability file delivered ready.
Unauthorized substitution is one of the few places where institutional maturity can actually be measured, since it tests not what a company did but its capacity to demonstrate what it did, and that capacity depends on who resolves the tension between the instinct to keep the line running and the discipline of keeping the record, and how quickly. An organisation that leaves this tension to the field engineer's initiative does so not because it doubts that initiative, but because it has never calibrated the approval channel to the tempo at which the field operates. The operative question is not whether substitutions occur, but who learns of them, when, and on the strength of which document.