Elite mountaineers don't decide whether to turn back when they're exhausted, oxygen-depleted, and within sight of the summit. They decide weeks before they start climbing. They set turnaround times: if they haven't reached a specific point by a specific hour, they descend regardless of proximity.

The rule exists because climbers know their judgment will be worst exactly when the decision matters most. When you're invested, exhausted, and close to a goal, you're the least qualified person to evaluate whether continuing is wise.

The same principle applies to strategic initiatives. The worst time to decide whether to kill a project is when you're deep in it, with sunk costs accumulated and identity attached. That's why you need to define your exit criteria before you reach that point.

Kill criteria aren't pessimism. They're decision architecture. You're protecting your future self from evaluating while cognitively and emotionally compromised by investment.

Why "In It" Is the Worst Time to Decide

Several predictable failures happen when you evaluate strategic exit while already invested:

Escalation via investment. As we covered earlier, the more you've put in, the harder it becomes to abandon. Barry Staw's research shows that feeling personally responsible for a decision makes you more likely to double down when it's failing. The sunk costs become psychological anchors.

Opportunity cost blindness. Every month spent on a low-expected-value initiative is a month stolen from higher-value alternatives. But when you're "in it," you're focused on the thing in front of you. The options you're not pursuing become invisible.

Negotiation posture. When signals appear that suggest killing the project, you negotiate with them. "Maybe the market will turn." "We're close to a breakthrough." "This quarter is an anomaly." The negotiation feels like prudence but often functions as delay.

Gary Klein's Premortem

Gary Klein, a research psychologist who studies decision-making in high-stakes environments, developed a technique called the premortem in the 1990s. The concept inverts traditional planning.

Instead of asking "what could go wrong?" before you start, you imagine the project has already failed and work backwards. The prompt: "It's twelve months from now. This initiative failed, or cost more than it was worth. Looking back, what were the early warning signals we ignored?"

This mental time travel produces different answers than standard risk analysis. When you project yourself into the future and assume failure has occurred, you're not defending the decision anymore. You're explaining a failure that already happened. The framing bypasses the optimism bias that makes operators underweight potential problems when they're excited about a plan.

Research has shown the premortem technique increases accurate risk identification by approximately 30% compared to standard approaches.

Premortem in Practice

Before committing to a major initiative, gather the team. Present this scenario:

"It's Q4 next year. We killed Project Atlas after investing $3M and eighteen months. Looking back, what were the signals that this wasn't going to work that we should have caught earlier?"

Have each person write independently for five minutes, then share. The independent writing prevents anchoring on the first person's ideas.

The signals generated become the foundation for your kill criteria.

Schelling on Precommitment

Thomas Schelling, the Nobel laureate in economics, made a counterintuitive argument: sometimes you're better off if your options are constrained in advance.

It sounds paradoxical. How can fewer choices be advantageous? But consider the mechanisms:

The general who burns the retreat route changes the incentive structure for the battle. The founder who announces a hard deadline creates external accountability. The climber who sets a turnaround time doesn't have to agonize on the mountain—the decision was already made.

Precommitment works because it separates the decision from the emotional context. You're not evaluating whether to exit while you're invested and close to the goal. You decided months ago what conditions would require exit. When those conditions appear, you're executing the protocol, not making a fresh judgment call.

"If you don't precommit, you'll negotiate with yourself forever. And the negotiation feels like hope. That's what makes it expensive."

The Kill Criteria Architecture

Kill criteria come in two forms:

Hard-stop criteria: If you observe this signal, you kill the initiative. No negotiation, no "let's see one more quarter." The signal triggers the exit.

Investigate-fast criteria: If you observe this signal, you take a specific information-gathering action within a defined timeframe (typically 2-4 weeks). Based on what you learn, you either continue for one more cycle or you kill it.

The distinction matters. Some signals are strong enough to warrant immediate action. Others require clarification. But both types are defined in advance, before you're compromised.

Kill Criteria Builder

Step 1: Define What You're Protecting

What's more important than this specific initiative succeeding?

These are your "getting home safely" goals. They matter more than any single summit.

Step 2: Run the Premortem

Have the team write independently for five minutes on:

"It's [timeframe] from now. This initiative failed or cost too much. Looking back, what were the early warning signals we ignored?"

Step 3: Sort Signals Into Buckets

A) Hard-stop criteria (if we observe this, we kill it):

B) Investigate-fast criteria (if we observe this, we take this action within 3 weeks):

Step 4: Set the Review Protocol

Review date: [specific date]

States that must be true by then: [measurable thresholds]

If not true: [specific action, typically kill or major redesign]

Failure Modes

Kill criteria systems fail in predictable ways. Knowing them helps you design better:

Criteria set too loose. "If revenue declines significantly" isn't a criterion. "If MRR falls below $50K for two consecutive months" is. Vague criteria enable negotiation.

No enforcement mechanism. Criteria without accountability are wishes. Who reviews? When? What's the decision process when signals appear? Without explicit governance, criteria drift into suggestions.

Criteria set retrospectively. Setting criteria after you're already invested is theater. The whole point is to make the decision before emotional attachment forms.

Single points of failure ignored. The monkey problem applies here. If the core hypothesis hasn't been validated, all other criteria are secondary. The most important kill criterion is often: "Has the foundational assumption been proven?"

The Governance Layer

Kill criteria need institutional backing to function. Without it, they become aspirational documents that get overridden when the moment arrives.

Consider:

The Portfolio Frame

Kill criteria become easier to set when you frame yourself as a portfolio manager rather than a project champion.

A portfolio manager expects some bets to fail. That's not a bug—it's built into the model. The goal isn't to make every investment work. It's to allocate capital such that the portfolio generates returns over time.

This frame changes the emotional weight of kill criteria. You're not setting up failure conditions. You're defining the rational rebalancing triggers for a portfolio of bets. Some will hit. Some will miss. The skill is identifying which is which as early as possible.

A killed initiative isn't failure. It's portfolio rebalancing. The skill is killing losers early enough to reallocate to winners.

Implementation for One Initiative

For one initiative you're currently uncertain about:

  1. Run a 10-minute premortem with whoever is closest to the decision
  2. Identify the 2-3 most likely failure signals that emerge
  3. Convert each into a specific, measurable criterion
  4. Designate a review date and decision owner
  5. Document in whatever format ensures it gets referenced

This isn't comprehensive governance. It's a minimum viable kill criteria system. Better than nothing. Iterate from there.

The series covered why operators over-persist (the persistence trap), why exit feels like failure (rigged scorecards), why teams work on the wrong problems (monkeys and pedestals), why the language itself creates drag (the euphemism tax), and now the structural solution: decide when to exit before you're in the fog.

The goal isn't to become a habitual quitter. It's to become a discriminating operator who can tell the difference between worthwhile persistence and expensive stubbornness. Kill criteria are the mechanism that enables that discrimination.

The Strategic Exit Series

Decision Architecture for Operators

This is Part 5 of a series on strategic decision-making under uncertainty.

← Previous: The Euphemism Tax
Start from the beginning →

This content is educational and does not constitute business, financial, or medical advice.