Asked in an investment committee session how demand was validated, a presenting team will typically answer with a list: the number of prospective customers spoken to, the institutions with which preliminary understandings were reached, the letters of intent currently on the table. The list is crowded, and its crowdedness effectively closes the committee's question. Yet most of the meetings on that list were occasions in which the product or the project was presented, the counterparty responded courteously in the affirmative, and the team recorded that response; the question put to the counterparty was whether the solution appealed, not which budget line would fund it and at what amount. The two categories of meeting sit side by side on the same calendar, land in the same CRM record, and are counted on the same slide, yet the information they generate differs structurally.
A second and less noticed pattern concerns the identity of the person interviewed. Within a corporate buyer, the purchase decision does not form at a single table; the unit that will use the product, the unit that carries the budget, the unit that grants technical approval and the unit with authority to defer the decision typically sit on separate reporting lines. Discovery conversations are almost invariably held with the most accessible and most willing counterpart — generally the unit that will use the product but does not carry the budget — precisely because that counterpart is available and speaks favourably. The budget holder's answer to the same question is constructed on a different logic: the renewal schedule of the incumbent solution, the internal allocation of resources, and that year's ranking of priorities. Where the second answer has not been obtained, what has been gathered is evidence of interest rather than evidence of demand.
This pattern has a name — **customer-discovery failure**, the absence of systematic testing of the demand assumption against an actual buyer — and its mechanics reflect a cost calculation rather than carelessness. Discovery is expensive; reaching the right counterpart takes weeks, interview calendars are hostage to corporate procurement cycles, and the information obtained frequently weakens the team's own hypothesis. Holding the assumption fixed, by contrast, is cheap and accelerates the process: the model is built, the budget is approved, the team begins work. The cognitive layer enters here, in that a person who sets out to test a hypothesis will search systematically harder for confirming evidence than for disconfirming evidence, and this asymmetry arises not from bad faith but from how the question was framed at the outset.
The shortcut deserves to be recognised as functional under specific conditions, since a remedy that ignores this will not survive contact with an operating business. In a market where the customer base is stable, where purchasing behaviour has repeated over years, and where the firm has accumulated hundreds of cycles in its own institutional memory, retesting the demand assumption from zero each time genuinely wastes resources; the assumption there is compressed historical data, and it performs adequately most of the time. The difficulty lies not in the shortcut itself but in the shortcut persisting after the condition that produced it has changed. Where a new geography, a new buyer segment, a product category whose regulatory frame has not yet settled, or a use case unfamiliar to the existing customer is involved, the distance widens between the information the historical data carries and the information the pending decision requires, while the confident tone of the assumption remains unchanged.
The most insidious form of this shift is the partial one. Where a new product is offered to the same customer, through the same sales team, along the same channel, but funded from a different budget line, the team skips discovery on the strength of the existing relationship; yet the moment the budget line changes, the decision architecture on the other side has changed with it, and the trust carried by the prior relationship is not legal tender in the new approval chain. In transitions of this kind, a sales cycle running to two or three times the initial estimate is a predictable outcome rather than a surprise.
Where the institutional cost accumulates is less visible than commonly supposed. An untested demand assumption does not open an immediate gap in the income statement; it accumulates first in the working capital cycle. Production or procurement planned against anticipated orders remains suspended in inventory above the prior year's level; receivable days lengthen as additional flexibility is extended to the customer; the sales team's conversion rate falls while the nominal size of the pipeline grows, since opportunities that fail to convert are not removed from the system. The simultaneous movement of these three indicators constitutes an early signal, readable months ahead of the revenue line, that the product has not found its market.
A second layer of cost surfaces at the moment the company changes hands or takes external capital. The way an acquirer examines revenue quality in diligence focuses less on the existence of revenue than on its repeatability: by what mechanism customer acquisition occurred, whether that mechanism can be reproduced independently of the founder's personal relationships, and on what assumption the current customer concentration rests. Where it emerges that the demand assumption was never tested in structured form, the consequence is usually not a reduction in the multiple but a change in the payment structure; a larger share of the consideration shifts into an earn-out, the escrow ratio rises, and the representations and warranties expand to cover the assignability of customer contracts. Incomplete discovery, in short, penalises the timing of cash flow before it penalises price.
In capital-intensive projects the same mechanism appears with a different vocabulary. There, the counterpart of customer discovery is the distance between an independent demand study and a binding offtake commitment: a market-size estimate in a feasibility report does not substitute for a purchase obligation supported by a counterparty signature, and credit committees price that distinction. Where the developer presents a portfolio of letters of intent as proof of demand while the lender reads the same portfolio as a conditional indicator, the resulting delay commonly adds one or two quarters to the financial close timetable, and the cost of that delay must be computed together with the carry on capital already burned during development.
This tendency is not managed through individual awareness; the person who skips discovery generally does so not by choice but because the process was designed that way. The neutralizing mechanism consists of four separable components. The first is fixing the **falsification threshold** in writing before the interviews begin: which answer, if received, will be accepted as having invalidated the assumption must be defined at the outset by the body that will take the decision, failing which any answer can later be read favourably. The second is a **buyer role map**, in which the user, the budget holder, the technical approver and the party with authority to defer are listed separately and a distinct interview target is set for each role. The third is the **separation between the party that formulates the hypothesis and the party that tests it**; having the test conversations conducted by someone under no obligation to defend the product changes the form of the question asked at its root. The fourth is the **evidence record**: who was spoken to, on what date, with which question, and the raw form of the answer received, held apart from interpretation.
Whether these components function depends on the record being kept at the moment of proposal rather than at the moment of approval. Customer evidence assembled after an investment decision has been taken is inevitably selected so as to support that decision; the same evidence, collected before the decision and against a written falsification threshold, produces information of a different order. Rhythm matters at least as much as content: rather than concentrating discovery conversations at project inception and then discontinuing them, running them as a cycle repeated at each step of capital commitment — the first budget, the scaling decision, the opening of a new segment — makes the ageing of the assumption visible.
BEIREK's intervention in this problem is framed not as advising teams to hold more conversations but as redesigning the decision gate. At every step leading to a capital commitment, we operate a record in which the single sentence carrying the demand assumption is written out explicitly, the observation that would invalidate that sentence is defined in advance, and a separate evidence entry is maintained for each of the four roles on the buyer side; this record is a precondition of the investment committee file rather than an annex to it. We separate interview design onto a line that does not defend the product, keep the raw answer apart from its interpretation, and tie the rhythm of assumption review to the drawdown points of capital rather than to the project calendar.
The strength of a demand assumption is measured not by the number of conversations that support it but by how serious an attempt was made to invalidate it and by the failure of that attempt; an untested assumption, however often it is repeated, is not evidence but an expectation shared inside the institution.