Sunk cost fallacy for projects

What it really means for project decisions

The pattern happens when past, irrecoverable investments drive continued commitment to a course of action, even when new evidence points to a poor outcome. In project terms, “we’ve already spent X” becomes the main justification for more time, budget, or scope rather than a reasoned reassessment of expected future returns.

Managers should treat the sunk cost argument as noise in the decision calculus: only future costs and benefits that can be influenced should matter for whether a project continues.

Why this keeps happening

• Psychological loss aversion: people hate admitting loss; continuing a project feels like avoiding loss.
• Identity and reputation: sponsors or project leads tie their competence to the project’s survival.
• Organisational signaling: cancelling looks like failure to peers or leadership, so teams delay.
• Structural inertia: budget cycles, contract terms, and handoffs create friction against stopping.
• Information gaps: early unevaluated commitments make later data hard to interpret.

These forces interact. For example, if a senior sponsor’s reputation is wrapped up in the initiative, team members will skew reports to avoid triggering a stop decision. Structural and psychological drivers reinforce each other, so the behaviour persists even when metrics clearly worsen.

How it shows up day-to-day

• Repeated “just give it one more sprint” requests despite declining KPIs.
• Adding scope, resources, or time instead of testing a reduced hypothesis.
• Defensive detail in status reports (data buried, optimistic framing) rather than straightforward tradeoffs.
• Reluctance to pause a pilot that misses targets because stopping requires admitting prior error.

These signs often begin small (a request for a short extension) and compound: each small extension becomes the new baseline for defending the next one. Left unchecked, the project accumulates commitment beyond what the updated business case supports.

A concrete workplace example

A product team had spent nine months building a new feature that analytics later showed had near-zero user engagement. The lead requested a full redesign and another three months, citing the investment so far. The product director paused funding and requested a two-week usability experiment and a pivot plan based on measurable user outcomes. The experiment confirmed low demand; the project was shut down and resources redirected to a higher-value backlog item.

That outcome required an explicit switch from defending past work to testing future value. Without the experiment and a neutral reviewer, the team would likely have continued with the redesign and sunk more costs.

• Sprint 1–6: Team builds complex dashboard.
• Sprint 7: Early metrics underperform.
• Usual reaction: add engineers, extend timeline.
• Better reaction: pause, run a trimmed prototype with real users for two sprints. If prototype fails, stop; if it succeeds, resume with clear milestones.

This mini-protocol creates a low-cost way to validate whether additional investment is justified.

Where it gets misread or confused

• Escalation of commitment: closely related. Escalation emphasizes increasing commitment despite negative outcomes; sunk cost is the cognitive justification (because we already invested). Escalation often points to organisational dynamics (power, politics) that drive continued investment.
• Optimism bias: belief that future outcomes will be better than warranted. Optimism can feed a sunk cost response but isn’t the same: optimism affects expectation-setting; sunk-cost reasoning uses past loss as the reason to continue.
• Confirmation bias: selectively noticing positive signals to support continuation. Teams may interpret weak data optimistically to avoid the uncomfortable decision of stopping.

Separating these helps choose remedies: governance changes can counter escalation of commitment, while decision rules and independent evidence reduce the influence of sunk-cost reasoning and confirmation bias.

Questions worth asking before approving more investment

• What specific future outcome are we buying with more time or money?
• Could we test the remaining risk with a smaller, time‑boxed experiment?
• Who would be affected by stopping, and how can we communicate the decision as learning?
• Are there contractual or structural barriers that make stopping needlessly costly?

Asking these keeps the focus on controllable, forward-looking value instead of prior losses.

Related patterns worth separating from it

• Planning fallacy: underestimating how long work takes. This can make sunk-cost situations more likely because timelines slip and teams keep adding time.
• Endowment effect: overvaluing something because you already possess it. In projects, teams overvalue work they invested effort in, making abandonment feel wasteful.

Understanding the distinct mechanics of each pattern helps tailor interventions: better estimation processes, clearer exit criteria, and decision roles mitigate different causes.