When an investment committee turns to a business line that has missed its targets for three consecutive periods, the discussion tends to organize itself not around the economics of the line but around the disposition of the people running it; the first half of the presentation quantifies the variance, while the second half explains how hard the team has worked, how slowly the market has matured, and how the inflection has slipped one quarter further out, and the decision is taken on the moral ground that second half establishes. Presented to the same committee for the first time, accompanied by today's performance data, that line would face a markedly low probability of approval, whereas the motion to continue it faces a high one — even though the question on the table is identical in both cases: how much capital and how much attention will be allocated to this line over the coming period.
The asymmetry surfaces in the language as well. Arguments for continuation are built from words like patience, discipline, long-horizon thinking and commitment, while arguments for termination are confined to a vocabulary carrying overtones of capitulation, retreat and abandonment; the choice between the two options therefore ceases to be a comparison of economically weighted alternatives and becomes an unequal selection in which one branch is tied to virtue and the other to failure. Agenda design reinforces the imbalance: the termination discussion typically falls at the end of the meeting, into the slot where time is short and the attention budget is exhausted, and it is deferred to the next session for want of time — deferral being, in practice, the continuation decision itself.
This pattern has a name — the perseverance trap, the continuation of a strategy that is not working on the strength of resolve rather than results. Its source is not a defect of reasoning; on the contrary, abandoning too early carries a serious cost of its own, since many business models run a negative cash curve at the outset, distribution channels mature late, reference accounts accumulate slowly, and enterprise sales cycles can span several budget years. Under precisely those conditions, the disposition to persist functions as a useful heuristic that prevents the value destruction premature closures would produce. The problem lies not in the heuristic but in its continuing to operate at full strength after the conditions that legitimized it have disappeared.
Three components of the mechanism reinforce one another. The first is sunk cost reasoning, in which the amount already spent, though unrecoverable, frames the decision in terms of cumulative investment looking backward rather than marginal return looking forward, so that stopping grows psychologically more expensive as the investment grows larger. The second is the fusion of decision and identity: to the extent that the professional standing of those who proposed, defended and managed the line is bound to its fate, adverse data about the line is read as a personal verdict, and the defensive reflex arrives ahead of the analytical one. The third is the ambiguity of feedback; a sufficiently noisy data set can sustain both the continuation narrative and the termination narrative at once, neither side can decisively falsify the other, and ambiguity resolves in favor of the status quo every time.
The single observable marker separating commitment from repetition is whether the thesis has been updated over time. In a sound continuation decision, the past period's data has changed the thesis — the target segment has narrowed, pricing has been restructured, the channel has shifted, or product scope has been cut — and the decision to proceed is built on that revised thesis. Where the same rationale recurs across three consecutive periods — the market has not matured yet, the inflection arrives next quarter, the team has just completed its reorganization — what is being sustained is not a strategy but the original belief about a strategy. From the moment learning stops, elapsed time ceases to be accumulated experience and becomes the same experiment run again.
The institutional cost of this disposition accrues less in the loss line than in the invisible allocation line. The most expensive input into a persisting business line is not the cash it consumes but the management attention it absorbs and the fact that the organization's most capable operating team cannot be deployed against an alternative; a second product line or a geographic expansion is delayed not for want of capital but for want of decision bandwidth. On the operating side the cost appears on more tangible surfaces: working capital requirements that grow as the sales cycle lengthens, inventory that stops turning, supplier commitments that cannot be unwound, and pricing discipline that deteriorates through low-margin work accepted to fill capacity.
At the valuation table, this disposition generates a signal heavier than the performance of any individual line. A buyer or a lender encountering lines that have gone years without closure, and without a revised thesis, records the observation not as a discrete investment error but as a structural read on the company's capital allocation discipline; the diligence question, accordingly, is rarely why the line was started, and almost always who determines the threshold at which it will be stopped and by what record that threshold is documented. Where no written answer exists, the transaction structure predictably hardens: earn-out scope widens, conditions precedent come to include closing or carving out the line, representations and warranties deepen, and the escrow ratio moves up.
The cost on the human side is slower to appear and more durable once it does. Teams that have given years to a line producing no result leave when they conclude that the internal path upward is blocked, and those who leave are typically the ones with the greatest capacity to find alternative employment; the cadre that remains consists disproportionately of the continuation decision's most loyal defenders, which further reduces the organization's capacity to generate a counter-argument at the next review. The cycle has a second-order effect as well: in organizations where a long-unstoppable line is finally stopped, the subsequent period tends to exhibit overcorrection, with new ventures terminated far earlier than their maturation curve would warrant.
This tendency is neutralized by decision architecture rather than individual resolve, and the architecture has four components. The first is recording the thesis at the moment of proposal rather than the moment of approval, so that the leading indicators and thresholds below which the thesis will count as falsified are written down while no one's reputation is yet attached. The second is removing continuation from the default position and reconstituting it as an active re-authorization, so that persistence occurs through a fresh allocation decision rather than through silence. The third is separating ownership of the termination assessment from ownership of the line, since redistributing the decision right structurally weakens the identity-fusion effect. The fourth is pricing the cost of exit at the outset; where supplier commitments, lease terms, personnel obligations and termination provisions in customer contracts have been structured so as to make stopping impractical, the freedom to decide was surrendered at the contracting stage.
The mechanism BEIREK builds into capital-intensive projects operates along exactly this line. At every stage gate — development, pre-FID, closing, first draw — a decision record is maintained defining which assumption will be treated as falsified at which threshold, and that record belongs to the proposal file rather than the approval file; what the subsequent review examines is therefore not performance in the abstract but the gap between the recorded assumption and the realized outcome. On the contracting side the same discipline is applied through staged commitment, priced optionality, symmetric construction of LD caps and termination provisions, and procurement commitments tied to project milestones — the objective being not to make termination easy but to make its cost knowable at the moment of decision.
The second layer is governance rhythm. In pre-mortem sessions run ahead of stage gates, the project is assumed to have failed and the reasons are written backward from that assumption; in those sessions the task of generating the counter-argument is not left to volunteerism but assigned to a defined role, and that role's output is appended to the committee file. Proposing a stop thereby ceases to be an act of personal courage and becomes an institutional function, with the person recommending termination positioned as someone performing the work expected of them rather than as the losing party. In agenda design, allocation decisions are placed at the opening of the session rather than the close, since what determines the quality of attention a decision receives is more often its position on the agenda than its substance.
The measure of institutional maturity lies as much in which undertakings are stopped, at what threshold and on whose authority, as in which were started; and the difference in record quality between those two acts speaks considerably louder, in the next financing round or the next valuation exercise, than any narrative the company tells about itself.