In an investment committee session, when two feasibility presentations prepared for the same project are set side by side, the divergence between them rarely lies in methodology; it concentrates instead in a single input — an occupancy assumption, a commissioning date, a terminal value multiple. Both presentations are internally coherent, both models are arithmetically clean, and both have selected a figure from a range that can be defended on its merits. What separates them is what the party choosing the number gains from that number. In the model prepared by the team developing the project, the commissioning date sits at the early end of the range; in the independent review it sits at the late end. Neither has misstated anything, both remain inside the same band of uncertainty, and yet the selection drifts systematically in one direction.
The same pattern repeats across supplier selection, advisor appointment, internal resource allocation and acquisition pricing. Where an advisor's compensation in a sale process is contingent on the transaction closing, the items appearing in that advisor's risk register are typically classified in a manner that does not delay closing — recorded as a matter to be remediated post-closing rather than elevated into a condition precedent. Where a business unit head's incentive is indexed to the scale of the unit's budget, the demand projections supporting that unit's investment cases sit persistently in the upper band. At no point in either sequence does an individual knowingly write something false; the output nevertheless behaves as though someone had.
This pattern carries a name — conflict-of-interest bias, the displacement of an evaluation's content by the evaluator's economic or institutional stake, independent of conscious intent. The defining characteristic of the mechanism is that intent is not required for it to operate. Every judgment rendered under uncertainty involves a selection from a range, and human judgment, in making that selection, carries a predictable tendency to move away from the end of the range that is costly to the person selecting. The evaluator finds the chosen figure defensible, and it genuinely is defensible; the only thing that is not defensible is that the same person, holding an interest pointing the other way, would very likely not have chosen it.
Under certain conditions this tendency is functional, and its complete elimination is not the objective. A development team's conviction in its own project is the engine of the development process itself; a team operating with no stake in the outcome does not chase permitting authorities, does not push a supplier negotiation to its final increment, and does not treat a site problem as its own. The classic tension in investment-readiness work sits precisely here: interest generates execution energy while the same interest degrades evaluative objectivity. The problem lies not in the presence of interest but in the convergence of the execution role and the evaluation role within a single individual or a single reporting line. When the condition changes — that is, when the moment of decision arrives — the tendency that had functioned as an engine becomes a force acting on the measuring instrument.
A second and less frequently recognized layer is that disclosure of the interest does not constitute a remedy. Institutional practice tends to treat a declaration of interest as a sufficient defence, whereas the observed effect of disclosure runs in two directions at once. An evaluator who has disclosed a stake regards the moral obligation as partially discharged and consequently states the recommendation with greater force; the receiving party, having heard the disclosure, knows that the advice warrants discounting but, lacking any basis for calibrating the magnitude, applies an insufficient correction in practice. Where the two effects compound, a disclosed conflict can generate a larger distortion than an undisclosed one. Disclosure is an instrument of transparency; it is not an instrument of control.
The institutional cost appears first in the schedule and only afterwards on the balance sheet. In a structure where evaluation and execution are not separated, adverse findings surface not in the early stages of the process but at the stage where reversal has become expensive — after a contract has been signed, after an advance payment has been released, after an equipment order has been fixed. The cause is not that the finding was discovered late; it is that the career cost of escalating a finding early appears higher than the cost of escalating it late. The consequence accumulates not in the capital expenditure line but in the timing of that expenditure — projects are not left unstopped, they are simply stopped two quarters later, and the commitments incurred across those two quarters cannot be unwound.
In a sale or investment process the cost translates directly into the language of valuation. Among the first tasks a buy-side diligence team undertakes is an examination not of the findings presented but of the compensation architecture of the party producing them: whether the technical advisor is paid on a fixed basis or on success, whether internal audit reports to the board or to the chief executive, whether the sales incentive is computed on invoiced revenue or on collected cash. Where these structures are found to be aligned with a directional outcome, the entire presented data set is discounted until independently verified item by item — and since item-by-item verification consumes time, the practical expression of that discount is a compression in the transaction multiple, an increase in the escrow proportion, or an expansion of the representation and warranty package. The company most often experiences this differential not as an independence problem but as evidence that the buyer is unduly cautious.
A third cost item is the degradation of institutional memory. Once assumptions shaped by interest are committed to writing, they become the reference point for subsequent decisions; a demand assumption selected at the optimistic end a year ago acquires, in the following year, the status of a figure the organization has accepted, and challenging it thereafter requires a separate justification of its own. In this way a displacement created in a single decision hardens into a baseline at the portfolio level. This is the most expensive form of conflict-of-interest bias, since its effect persists inside the model long after its source has disappeared — after the relevant manager has left, after the incentive structure has been redesigned.
The first component of a structural intervention is not disclosure of the interest but separation of authority: the party proposing an assumption and the party approving it should sit in different reporting lines, and the approver's economic outcome should be tied to the accuracy of the decision rather than to its direction. The second component is advancing the moment of record — the decision record is kept not in the approval meeting but at the instant the assumption first enters the model, capturing who proposed it, from which range it was drawn, and why the opposing end of that range was rejected. The third component is the formal assignment of the counter-argument as a role: a party that incurs no cost from the project's rejection, and is expected to produce contrary findings, is appointed at the outset of the process on compensation terms unrelated to the outcome. The fourth component is documentation of the compensation architecture itself; where it stands in writing which role gains what from which outcome, an external review reads that document not as an exposure but as evidence of control.
The operating structure BEIREK applies in capital-intensive projects establishes this separation in practice. While carrying project management responsibility across development, engineering and contracting lines, the review of the assumption set is conducted on a separate line under a distinct record discipline; for each critical input, the value selected is recorded alongside the rejected end of the range and the rationale for that rejection, and the resulting record is submitted to the investment committee as an annex to the model. The element that proves most useful in application is that this record is initiated not at FID but at the first technical assumption session preceding the term sheet — because the displacement is formed in that session, whereas the decision session merely institutionalizes it.
A second line of intervention carries interest alignment on the procurement and contractor side into the contract itself. In EPC and advisory agreements, the milestone to which fees are attached, the level at which the liquidated damages cap is set, and the party holding technical approval authority together determine, to a substantial degree and well in advance, which findings will surface at later stages of the project; these clauses are therefore negotiated not as elements of commercial bargaining but as components of decision quality. In the post-closing period a review cadence of fixed rhythm is maintained: assumptions are compared against realized outcomes, the input in which the deviation concentrates is tracked, and the opening range for the next project is calibrated against that deviation.
The single indicator of whether an organization can manage conflict of interest is not the existence of its ethics declarations but its answer to a narrower question — by whom, and at what stage, an assumption unfavourable to its own position can be entered into the record. Where a structure cannot answer that question precisely, decision quality is read not from the competence of the analytical team but from whose interest happened to point in which direction during the period under review, and this is the first thing any external examination notices.