The material placed on an investment committee table as evidence of demand tends to consist of the same three exhibits: the order curve for the first two or three quarters, a folder of signed letters of intent, and favourable feedback from pilot accounts. What is rarely stated in the same session is that none of the three contains a second purchase decision. The curve counts first orders only; the letter of intent carries no budget commitment; the pilot feedback measures the satisfaction of a user spending from an evaluation allowance. The pattern observed across sectors is consistent enough to be treated as structural: there is a systematic divergence between early order density and the repeat-purchase rate of the same product eighteen months later. That divergence typically becomes visible only after the capacity decision has been taken, the supplier commitment signed and the team expanded — which is to say, the measurement arrives outside the window in which it could still correct the decision.
The same pattern can be read directly in the architecture of the weekly sales report. Such a report ordinarily carries new customer count, average order value and the number of open opportunities in the pipeline; it does not carry the share of revenue from repeat buyers, the proportion of accounts purchasing in two consecutive periods, or the elapsed time between first and second order. The reason is ownership rather than oversight: the first three figures measure the performance of the sales function and sit squarely within its remit, so they are produced every week without prompting, whereas the latter three sit at the intersection of product, operations and finance and therefore appear on no single agenda. A metric without an owner does not enter the report, and a metric absent from the report is treated as nonexistent at the moment of decision. The distortion in the demand reading thus originates not as an error of judgment but as an artefact of reporting architecture.
This reading has a name — market-pull misreading, the interpretation of a transient signal of market pull as a durable demand structure — and in the early phase it functions largely as a rational shortcut. The observation window required to establish whether a product has genuinely found its market is longer than the cash window available to most young companies, so treating the only available early signal as proof of demand is a defensible way to proceed under resource constraint. The difficulty lies not in the shortcut itself but in its persistence after the underlying condition has changed: once the company reaches a scale at which it can commit capital, the same signal ceases to serve as a discovery indicator and begins to serve as a justification for commitment. Because that transition carries no explicit threshold, the decision architecture ordinarily continues to operate on the earlier assumption.
The criterion that most reliably separates the two kinds of signal is the budget line from which the first order was funded. In a new category, early buyers typically spend from trial, innovation or project allowances — lines that carry comparatively low approval thresholds, can be released on the initiative of a single manager, and lapse if unspent within the year. A repeat purchase, by contrast, requires migration into the operating budget, and that migration involves a different approval chain, a formal vendor registration process, and in many institutions a second-source requirement. The first order therefore measures the curiosity of an individual; the second measures the dependence of an institution, and the gap between the two measurements has less to do with the quality of the product than with the internal mechanics of the buying organisation. The magnitude of the early wave carries almost no information about whether that migration will occur.
The conditions that manufacture a transient pull signal are, for the most part, identifiable in advance. Founder-led selling establishes a relationship of personal trust on the buyer side rather than an institutional supplier relationship, and personal trust does not survive a change in the buyer's own personnel. Introductory pricing depresses the decision threshold artificially by bringing the buyer's cost of evaluation close to zero. Periods of supply scarcity route buyers toward suppliers they would not otherwise select, and that routing reverses when scarcity abates. The first year in which a regulatory obligation takes effect produces a comparable one-off wave; once the compliance requirement has been satisfied, there may be no structural reason for the same buyer to purchase at equivalent volume in the following year. Each of these conditions generates genuine revenue; none of them generates repeatable revenue.
The first tangible cost of the misreading surfaces in the working capital cycle. Inventory calibrated to the early wave, combined with minimum order quantity commitments on the supplier side, slows inventory turns markedly once demand normalises, locking cash inside unsold goods. Line items such as tooling, production line setup or purpose-built equipment push unit cost upward because their amortisation rests on a volume assumption; when the volume fails to materialise, the excess cannot be passed into price and lands in margin instead. Headcount added and leases signed in the same period unwind far more slowly than inventory does: severance obligations and the remaining term of a lease continue to carry the fixed-cost base irrespective of any correction in demand. The company ends up paying, across several periods, for a signal misread in one.
The second cost appears when the company arrives at a capital raise or a share transfer. The analyst on the review side looks not at the total of the income statement but at the cohort distribution beneath it: how many of the customers acquired in the first quarter were still purchasing in the fourth, what share of revenue derives from first orders, what proportion of the total sits with the five largest accounts. That these tables have never previously been produced internally is often more determinative than whatever they reveal, because when the question asked at the diligence table is a question the company has never asked itself, what is priced is not only the answer but the maturity of governance. At that point the valuation multiple becomes sensitive not to the level of revenue but to the demonstrability of revenue that repeats independently of the founder and of one-off conditions.
That sensitivity translates directly into transaction structure. Where repeat-purchase data is thin, the buy side typically declines to reduce headline price and instead proposes structures that distribute the risk across time: payment of a portion of consideration through an earn-out tied to repeat revenue in future periods, a closing condition requiring binding framework agreements with named customers, an expansion of representations and warranties to cover the customer contract base, and an escrow ratio set above the customary band. A comparable reflex appears on the debt side, where the credit committee builds its base case not on first-year revenue but on the portion of revenue demonstrably capable of repetition, with the difference finding expression in a covenant heading or in the scope of security. Uncertainty in the demand reading is thus priced not as a topic of discussion but as a contractual term.
The mechanism that neutralises this tendency is decision architecture rather than individual caution, and it separates into four components. The first is classification of demand evidence by degree of bindingness: verbal interest, a letter of intent, a paid pilot, a framework agreement containing minimum volume, and an actual repeat order all belong in the same table but carry different weight, and committee material should state which rung each figure came from. The second is a cohort ledger maintained on a monthly cadence under a single named owner, living as a document distinct from the sales performance report. The third is classification of capacity decisions by reversibility — contract manufacturing against owning a line, fixed-term engagement against permanent headcount, short lease against long commitment — which creates the option of answering the same demand signal at different levels of commitment. The fourth is a counter-thesis role, charged with arguing the case in which demand recedes, held by someone other than the owner of the decision.
In capital-intensive and financed projects, BEIREK installs this mechanism as part of the project management discipline rather than as an analytical afterthought. Demand evidence enters the investment decision file not as free text but as a record carrying fields for bindingness rung, counterparty, volume commitment, term and termination condition, so that every volume assumption reaching the committee can be traced individually to its source. Capacity, equipment and employment line items are tagged by reversibility class, and no commitment proceeds to approval without being matched against the bindingness rung of the demand signal that supports it. Minimum volume obligations on the supplier side are indexed, so far as commercially achievable, to binding commitments on the customer side; where the gap between the two cannot be closed, it sits in the file as an explicitly priced position rather than as a buried assumption.
On the operating side the same discipline is bound to a cadence. The cohort table, the conversion rate from first order to second, and the elapsed time of that conversion form a standing agenda item in the monthly project review, reported by an owner independent of the sales function. Before the investment decision, a pre-mortem session is run on the assumption that the observed demand arose from a single non-recurring condition; the output of that session is not an opinion but a trigger list defining the thresholds at which the capacity decision is to be halted. In the post-closing period those same triggers become the decision points governing the staging of commitment, so that expansion investment is released not by a single approval but by repeat-purchase data crossing a defined threshold.
The difficulty of reading an early demand signal arises not from the signal being false but from the question it answers differing from the question required at the moment of decision: a first order establishes that a product is considered worth trying, while a second establishes that an institution has become dependent on it, and capacity can only be priced against the latter. The question an investment decision file ought to pose is therefore not how much demand has grown, but how much of it would remain standing once a particular condition is removed.