In 1985, Hal Arkes and Catherine Blumer presented participants with a simple scenario: you have purchased a $100 ticket to a ski trip in Michigan and a $50 ticket to a ski trip in Wisconsin. You then discover both trips fall on the same weekend. You can only attend one. Which do you choose?

The majority chose Michigan — not because they expected it to be more enjoyable, but because they had spent more money on it. They were optimizing for past expenditure, not future value.

This is the sunk cost fallacy. It operates in businesses of every size, every day, destroying capital, consuming time, and preventing decisions that the evidence has already made obvious.

A software company spent $1.4 million developing a product feature over eighteen months. User testing showed minimal adoption. The feature addressed a problem that market research — conducted after development began — revealed most users did not have. The engineering team recommended discontinuing the feature. The CEO continued funding it for another nine months.

When asked why, the CEO's answer was instinctive and immediate: "We've already invested $1.4 million. We can't just walk away from that."

The $1.4 million was gone whether the feature continued or not. The only relevant question was: will the next dollar spent produce a return? The data said no. The sunk cost said keep going. The sunk cost won.

This pattern repeats across every business function. Marketing campaigns that aren't converting but continue because the creative was expensive to produce. Employees who aren't performing but stay because the company invested heavily in their recruitment and training. Office leases that are too large for current needs but are maintained because breaking the lease feels like admitting a mistake.

In each case, the logic is identical: past spending justifies future spending. And in each case, the logic is wrong.

The sunk cost fallacy is not a failure of intelligence. It is a feature of human psychology. Kahneman and Tversky's prospect theory explains part of it: losses feel roughly twice as painful as equivalent gains feel pleasurable. Abandoning a project represents a crystallized loss — the acknowledgment that money was spent for nothing. Continuing the project maintains the possibility, however slim, that the investment will eventually pay off.

The brain prefers the uncertainty of continued investment to the certainty of a recognized loss. This preference is irrational but deeply felt. It explains why founders will fund failing ventures for years past the point where the evidence became conclusive.

There is also an identity component. For the person who championed the project, approved the budget, or hired the employee, discontinuation is personal. It says: I was wrong. In organizations where being wrong carries professional consequences, the incentive to maintain failing commitments is structural, not just psychological.

The most effective defense against sunk cost is to establish kill criteria before the investment begins. Not goals. Not success metrics. Kill criteria — the specific conditions under which the project will be terminated, the hire will be let go, or the strategy will be abandoned.

"If the product does not achieve 500 active users within 90 days of launch, we will discontinue development." That sentence, written before a dollar is spent, removes the sunk cost from the decision. When day 90 arrives and active users are at 200, the decision has already been made. The emotional weight of $1.4 million in development costs does not enter the equation — because the kill criteria were set when the sunk cost was zero.

This approach requires discipline and honesty. It requires writing down the criteria, sharing them with stakeholders, and genuinely committing to acting on them. Most importantly, it requires a culture where terminating a project is not treated as failure but as evidence of good judgment.

When kill criteria do not exist — which is most of the time — there is a simpler diagnostic. Ask: if I had not already made this investment, would I make it today?

If you had not already hired this employee, would you hire them today based on their current performance? If the answer is no, the decision is clear — regardless of what you spent on recruitment.

If you had not already built this feature, would you begin building it today based on current user data? If the answer is no, continuing is not prudent. It is compulsive.

If you had not already signed this lease, would you sign it today based on current space needs? If the answer is no, exploring the exit — even at a cost — may be the rational choice.

The replacement test strips away the emotional weight of past spending and forces a forward-looking evaluation. It does not guarantee the decision will be easy. It guarantees the decision will be made on the right basis.

Every dollar you have already spent is gone. Every dollar you spend next is a choice. The sunk cost fallacy is the failure to distinguish between the two — and it is the most expensive psychological error in business.