The Sunk Cost Fallacy: How to Spot a Failing Initiative Before It's Too Late

In May 2026, the US Army Corps of Engineers terminated the contractor on the Chickamauga Lock replacement project, citing sustained execution failures and missed schedules, and announced it would re-bid the work. The official statement described the decision as necessary "to protect taxpayer dollars." What was less visible in the announcement was the timeline: by the time the termination was executed, the project had been demonstrably failing against its key milestones for an extended period.

The decision to stop came significantly after the evidence that stopping was necessary. This is not an anomaly of government contracting. It is the predictable output of every organization whose strategic governance does not include structured off-ramps.

Megaprojects and major corporate initiatives rarely fail overnight. They decay slowly, in plain sight, while leadership finds reasons to continue. The reason is not ignorance. The people closest to failing initiatives almost always know they are failing before the formal governance structure acknowledges it. The reason is the sunk cost fallacy — the cognitive and organizational mechanism by which prior investment becomes the primary argument for continued investment, regardless of the expected future return.

Why organizations continue funding what they know is failing

Barry Staw's foundational research on escalation of commitment established that the primary driver of continued investment in failing initiatives is not ignorance of the failure but the psychological cost of acknowledging it. Leaders who championed an initiative have staked their credibility on its success. The organizational culture's treatment of stopping — whether it is coded as failure or as intelligent governance — determines whether that acknowledgment can happen before the damage becomes existential.

In cultures where stopping is treated as failure, the incentive structure runs in precisely the wrong direction. The person who surfaces the evidence that an initiative should be terminated takes on the reputational risk of the termination while the people who have been enabling the failure for months absorb no comparable cost. The rational response, in those cultures, is to continue optimistic reporting until the failure becomes undeniable — by which point the additional cost incurred during the period of denial is sunk, and the reputational damage is larger than it would have been had the evidence been surfaced earlier.

The process failure that enables the sunk cost trap

The sunk cost trap is not primarily a psychological problem. It is a process design problem. In organizations whose strategic governance is structured around reporting on execution progress — where the question in every review is "how is this going?" rather than "should this continue?" — the governance process never creates the conditions for the second question to receive a genuine answer.

The initiative is presumed to be continuing. The review assesses whether it is continuing well. The structural condition required to stop it — a formal re-evaluation of whether the strategic logic that justified the investment still holds — is never triggered because it was never designed into the process.

Effective strategic governance pre-commits to specific conditions under which re-evaluation is triggered. Before an initiative launches, the governance structure defines what failure looks like in measurable terms — not "the initiative is not progressing as hoped" but "if milestone X is not achieved by date Y, this initiative will be formally re-evaluated against the alternative uses of the resources it is consuming." This pre-commitment separates the governance question from the ego investment.

The decision to re-evaluate is not a fresh judgment by the people who championed the initiative about whether to acknowledge their own failure. It is the execution of a commitment made before the investment of time and reputation that makes acknowledgment so costly.

The early warning signals that organizations ignore

The signals of terminal initiative decline are almost always visible before the quantitative evidence is undeniable, and they are almost always process signals rather than outcome signals. The strategic review stops asking whether the initiative should continue and focuses entirely on how to improve execution — the question of continuation has been removed from the agenda.

The performance metrics are revised to make current performance look more acceptable against the revised baseline. The people closest to the initiative stop raising concerns in formal forums while continuing to raise them in informal conversations. The reporting becomes systematically more optimistic than the ground-level assessment of the people doing the work.

Each of these signals is an indicator that the sunk cost logic has taken hold of the governance process — that the social dynamics of the review have organized themselves around continuing rather than honestly evaluating.

The pre-mortem, conducted before the initiative launches, produces the most reliable antidote: a written record of the specific conditions that would constitute failure, agreed by the people who will later be responsible for reporting progress, before they have any ego investment in the reporting being optimistic. That record does not prevent the sunk cost trap from forming. But it provides the governance structure with a mechanism for triggering the re-evaluation that the trap would otherwise prevent.