Sunk costs at the level of strategy

Strategy & Decisions2026-10-026 min readAtlas Strategy Group

At the project level, the sunk-cost trap is familiar and the cure is known: money already spent has no vote in the decision about spending more. Anyone who has sat in that meeting knows the cure works imperfectly — but it at least exists, because project costs are visible, denominated in currency, and easy to argue about.

At the level of strategy, the same trap operates with different bait, and the standard cure quietly stops working. What is sunk in a strategy is rarely money. It is the market position the company has announced publicly. The promises made to investors, partners and staff. Two years of hires aligned to the chosen direction, and the founders' own recorded conviction — in interviews, in decks, in the story the company tells about itself. These investments are real and unrecoverable, and they appear on no line item, which is exactly why people who would never continue a dying project continue a strategy they privately no longer believe in. The accounting department cannot flag what does not appear in the accounts.

Why escalation is stronger here

The mechanism that keeps a bad strategy alive is audience. Changing a project is a private act: a meeting, a decision, a close-out note. Changing a strategy is a public one — and the perceived cost of the turn grows with the number of people who have heard the commitment. Once that cost feels higher than the cost of drifting, the company starts optimizing for looking right rather than being right: interpreting ambiguous signals in the strategy's favor, funding the direction just enough to keep the story alive, treating each quarter's disappointment as a reason to recommit. Consistency stops being a virtue and becomes the objective. It is a special kind of prison: the walls are made of the company's own speeches.

What actually helps

• Triggers written at adoption, not at doubt. The moment to define what would make the company change course is the moment the strategy is chosen — when nobody's credibility depends on the answer. A strategy with written revisit triggers can be reviewed against its own terms; a strategy without them can only be abandoned in crisis, at the worst possible price. This is the same discipline that separates decision quality from outcome quality: the standard is set before the evidence arrives.
• Theses instead of identity. A strategy stated as a testable thesis — «we believe this market rewards this positioning, and here is what would show we are wrong» — can be disproven without anyone losing face. A strategy stated as self-definition («we are an X company») can only be disproven by identity crisis. The grammar of the strategy document determines how cheaply it can later be revised.
• The entry test, asked honestly. One question, asked by the owner on a fixed schedule: knowing what we know now — the market, the results, our own capacity — would we enter this strategy today, at this price? The question is powerful precisely because it removes history from the calculation, and history is where sunk costs live. Under uncertainty, the habit of re-asking it on evidence rather than on anniversary is the practical core of strategic decision-making.

Sunk costs at the project level argue from the budget. At the strategy level they argue from identity — and identity negotiates much harder.

The uncomfortable summary: a strategy should be treated as a renewable commitment, reviewable against declared evidence, with its exit designed at entry — exactly the way healthy companies treat projects. The companies that manage this keep their reversals cheap and their credibility intact; the ones that don't pay for the drift quietly, until the market charges the full price at once.