In September 2008, as Lehman Brothers collapsed and the global financial system teetered on the brink, Jamie Dimon, CEO of JPMorgan Chase, made a series of decisions that would define his leadership legacy. Within 72 hours, he decided to acquire Bear Stearns' remnants at a steep discount, committed $30 billion of the bank's capital to stabilize markets, and simultaneously tightened lending standards to protect the bank's core business. Each decision involved massive uncertainty, incomplete information, and irreversible consequences. Years later, when asked how he made decisions under such extreme pressure, Dimon's answer was instructive: "You never have all the information. You have to be comfortable deciding with what you have, and you have to be clear about what kind of decision you're making."

The distinction between types of decisions — and knowing which approach each demands — is the foundation of leadership decision-making. Leaders who apply the same process to every decision are either paralyzed on routine choices or reckless on consequential ones. The discipline is calibrating the approach to the stakes.

The Cost of Indecision

Before examining how to decide, it is worth understanding the cost of not deciding — because leadership cultures that reward caution over decisiveness create enormous, invisible organizational costs.

The economic cost: Every day a significant decision is delayed, resources remain allocated to the current state rather than the chosen direction. Projects cannot start. Problems cannot be resolved. Teams cannot align. The daily economic cost of indecision on significant organizational choices is real and measurable.

The morale cost: Teams that await leadership decisions experience a specific form of demoralization: they have clarity about the problem but not about the solution, which means they cannot act effectively. This uncertainty is particularly corrosive for high-performers who are most capable of acting and most frustrated when blocked from doing so.

The signal cost: Every delayed decision sends an organizational signal about priority, urgency, and leadership capacity. Teams infer from decision delays that the decision is less important than it appeared, that leadership is unable or unwilling to engage, or that the organizational culture punishes commitment. These inferences are often wrong — but they are predictable consequences of indecision.

Example: General Electric under Jack Welch was famous for decision velocity. Welch believed that the cost of a wrong decision that was corrected quickly was lower than the cost of a slow decision process that reached the right answer too late. His famous "no. 1 or no. 2 or fix, sell, or close" policy was applied with unusual speed precisely because he believed that the decisive action — even an imperfect one — was better than extended uncertainty.

Four Decision Types: Approach Guide

Decision Type Stakes Reversibility Recommended Approach Common Failure Mode
Routine operational Low High Delegate or decide quickly with 70% of ideal information Over-analyzing; wasting cognitive resources
Significant reversible Medium Moderate Gather broad input; set deadline; decide and own it Seeking consensus instead of input
Significant irreversible High Low Run pre-mortem; pressure-test assumptions; seek dissent Indefinite analysis to delay discomfort
Crisis Varies Often low Decide fast with available information; communicate immediately Freezing or waiting for certainty that will not come

"Speed matters in business. Many decisions and actions are reversible and do not need extensive study. We value calculated risk taking." — Amazon Leadership Principles

Four Types of Decisions Requiring Different Approaches

Type 1: Routine Operational Decisions

Characteristics: Reversible, frequent, bounded, within established operating parameters.

Examples: Hiring a specific candidate among three qualified finalists, approving a vendor invoice, prioritizing the sprint backlog, selecting a meeting time.

Approach: Delegate as far as possible. For decisions that remain with leadership, decide quickly using heuristics and judgment. Apply the 70% rule: if you have 70% of the information you would ideally want, decide. The remaining 30% of information would have required time and resources disproportionate to the decision's stakes.

The failure mode: Over-analyzing routine decisions. Leaders who apply elaborate frameworks to routine choices waste cognitive resources, slow the organization, and signal to their teams that they do not trust their judgment or the established parameters.

Type 2: Significant Reversible Decisions

Characteristics: Significant but reversible, moderate stakes, affects multiple people or functions.

Examples: Organizational restructuring within your purview, product feature prioritization, process changes, vendor selection for non-critical systems.

Approach: Gather sufficient input from affected stakeholders. Set a decision deadline and honor it. Communicate the decision and reasoning clearly. Build in a review point after implementation.

The key discipline: Gathering input is not the same as seeking consensus. Input improves decision quality. Consensus is often impossible on significant decisions and produces compromise decisions that satisfy no one fully. Gather broad input, decide based on what you heard and your own judgment, and own the decision.

Example: Amazon's leadership principle "Bias for Action" explicitly values speed over perfection: "Speed matters in business. Many decisions and actions are reversible and do not need extensive study. We value calculated risk taking." This principle is calibrated to the type of decision — it does not apply to architectural decisions for mission-critical systems, but it does apply to marketing experiments, product features, and process improvements.

Type 3: Significant Irreversible Decisions

Characteristics: High stakes, difficult or impossible to reverse, affects many people over a long period.

Examples: Major acquisitions, strategic pivots, capital investments, leadership team changes at senior levels.

Approach: This is where extended analysis, broad consultation, and deliberate process are genuinely warranted. Run a pre-mortem: "Imagine it's two years from now and this decision was a disaster. What happened?" Pressure-test assumptions explicitly. Seek out dissenting views and engage them seriously rather than dismissing them.

The key discipline: Even for irreversible decisions, there is a point of diminishing returns on analysis. More information reduces uncertainty but never eliminates it. The decision timeline should be set and kept — delay does not eliminate risk, it just allocates it differently.

Type 4: Crisis Decisions

Characteristics: High urgency, incomplete information, no opportunity for extensive consultation, high stakes.

Examples: Responding to a system outage, managing a regulatory emergency, addressing a leadership crisis, responding to a major competitive threat.

Approach: Stabilize first, then analyze. The immediate objective in any crisis is to stop the bleeding — to prevent the situation from deteriorating further while you gather information. Decisive early action, even if imperfect, is almost always better than waiting for complete information that will never fully arrive.

Example: When Johnson & Johnson faced the Tylenol poisoning crisis in 1982, CEO James Burke made a rapid, irreversible decision: recall 31 million bottles of Tylenol from store shelves, at a cost of $100 million, before the full scope of the crisis was understood. The decision was made on incomplete information, under time pressure, with potentially enormous consequences in every direction. The decision to act decisively and protectively was judged, in retrospect, as one of the best crisis management decisions in corporate history. Burke did not wait for complete information; he acted on values.

The 70% Rule

Jeff Bezos articulated one of the most practical guidelines for decision-making: "Most decisions should probably be made with somewhere around 70% of the information you wish you had. If you wait for 90%, in most cases, you're probably being slow."

The 70% rule is not a precision formula — it is a heuristic for calibrating the relationship between information completeness and decision timing. Its value is in naming the implicit trade-off: additional information has diminishing returns, and the cost of gathering that information (time, resources, opportunity cost) may exceed its decision-improvement value.

When the 70% rule applies:

  • For reversible decisions where wrong answers can be corrected
  • For decisions in fast-moving competitive environments where timing is strategic
  • For decisions where the costs of delay (in morale, momentum, and opportunity) are significant

When to exceed the 70% threshold:

  • For irreversible decisions with major long-term consequences
  • For decisions with ethical dimensions where mistakes are not merely costs but wrongs
  • For decisions that establish precedents that will constrain future choices

Building and Using Decision-Making Frameworks

The Decision Journal

Annie Duke, professional decision scientist, recommends keeping a decision journal: a record of significant decisions, the reasoning behind them, and the expected outcomes — written before the outcome is known. The journal serves two functions.

First, it creates accountability: when you document your reasoning before the outcome is known, you cannot revise your logic afterward to make your decisions look better in retrospect than they were. This combats hindsight bias.

Second, it enables calibration: over time, you can assess the quality of your decision process by comparing expected outcomes to actual outcomes. If your decisions consistently produce different outcomes than you predicted, your models need updating.

The Pre-Mortem

Developed by psychologist Gary Klein, the pre-mortem is the single most powerful tool for improving decisions before they are made. The process:

  1. Gather the decision-making group
  2. State the decision that is about to be made
  3. Ask everyone to imagine that it is one year from now and the decision has turned out to be a complete failure
  4. Have each person write down, individually, what they believe caused the failure
  5. Go around the room and collect all the failure scenarios
  6. Use the failure scenarios to strengthen the decision: address the most likely failure modes, adjust the approach, or recognize that the decision is genuinely too risky given what has been surfaced

The pre-mortem works because it gives people permission to surface concerns they might otherwise suppress. In a normal decision discussion, raising concerns can feel like disloyalty or negativity. In a pre-mortem, everyone is explicitly tasked with finding failure modes, which makes concern-raising not just acceptable but expected.

The Regret Minimization Framework

For genuinely uncertain, high-stakes decisions, Jeff Bezos's regret minimization framework offers a long-horizon perspective: project yourself to age 80 and ask which choice you would regret more. This framework is particularly useful when short-term pain is preventing a decision that makes obvious sense over a long horizon.

Making Decisions Through Others: Delegation

The most consequential decision-making skill for leaders is knowing which decisions to make themselves and which to delegate. The failure modes are symmetric and equally damaging:

Under-delegation (making decisions that should be delegated):

  • Slows the organization because decisions queue at the leader's bottleneck
  • Demotivates capable team members by signaling distrust
  • Consumes leadership cognitive resources that should be directed at higher-stakes decisions
  • Prevents the development of decision-making capability in the team

Over-delegation (delegating decisions that should be made by leadership):

  • Creates decisions without appropriate accountability
  • Produces decisions that conflict with organizational strategy or values
  • Leaves team members exposed to decisions they do not have the authority or information to make well

The delegation decision rule: Delegate all decisions that (1) fall within the scope of someone else's authority and expertise, (2) do not require information only you possess, and (3) do not set organizational precedents or involve the organization's core strategic or ethical commitments. Decisions that touch on ethics require a different kind of care: applying a structured approach to ethical decision making rather than relying purely on intuition or expediency.

Common Decision-Making Failure Modes in Leaders

Decision Avoidance

The pattern: Framing decisions as "under discussion" or "requiring more information" indefinitely, without reaching a decision.

The cause: Fear of being wrong. In organizations that punish mistakes, the safest strategy for a leader is to delay decisions until circumstances make the decision obvious — or until someone else makes it.

The cure: Recognize that delayed decisions have costs that are as real as wrong decisions. Set explicit decision deadlines and honor them.

Premature Closure

The pattern: Settling on an answer quickly, then filtering subsequent information to confirm the chosen answer.

The cause: Cognitive efficiency. The brain dislikes ambiguity and will produce premature certainty to resolve it. Confirmation bias then reinforces the premature choice.

The cure: Actively seek disconfirming evidence. Assign someone the explicit role of devil's advocate. Run the pre-mortem. Ask "what would change my mind?"

Groupthink

The pattern: The group converges on a decision not because it is optimal but because the social dynamics suppress dissent.

The cause: Authority gradients, desire for harmony, and the social cost of being the person who says "I don't think this is right."

The cure: Create structures that explicitly invite dissent. The pre-mortem is one. Anonymous input mechanisms are another. Explicitly separating idea generation from idea evaluation prevents premature convergence.

Sunk Cost Fallacy

The pattern: Continuing a failing course of action because of resources already invested.

The cause: The emotional difficulty of admitting a mistake, compounded by social and organizational dynamics that punish reversal as inconsistency.

The cure: Evaluate decisions on prospective returns, not retrospective investment. The key question: "If we were starting fresh today with no prior investment, would we make this same decision?" If no, the sunk cost is irrelevant to the forward decision.

For related frameworks on how decision-making works in tradeoff contexts, see thinking in tradeoffs.

The Research on Decision Quality in Organizations

The academic study of how leaders actually make decisions -- as opposed to how they are advised to make them -- has produced findings that are both humbling and actionable.

Daniel Kahneman of Princeton University, whose 2011 book Thinking, Fast and Slow synthesized decades of research conducted with Amos Tversky, documented systematic biases that affect even trained decision-makers. In a study with experienced clinical psychologists, professional auditors, and military officers, Kahneman and colleagues found that these experts showed lower internal consistency in their judgments than their own confidence suggested: when given identical cases at different times, experts contradicted their own previous judgments roughly 40-60% of the time -- a finding Kahneman termed "noise" in a 2016 Harvard Business Review article co-authored with Andrew Rosenfield, Linnea Gandhi, and *Tom Blaser. The organizational implication is that individual leader judgment, however experienced, is less reliable than leaders believe. Structured decision processes -- checklists, explicit criteria, documented reasoning -- consistently outperform unstructured expert intuition in domains from medical diagnosis to personnel selection to investment analysis.

Philip Tetlock of the University of Pennsylvania conducted a 20-year study of expert political judgment, published as Expert Political Judgment (2005), tracking 82,361 probability judgments from 284 experts across political, economic, and strategic domains. His finding, which became influential across leadership and strategy contexts, was that average expert judgment was barely better than chance for long-range predictions -- but that this average obscured a meaningful variance. "Foxes" (experts who incorporated many sources of information and maintained calibrated uncertainty) significantly outperformed "hedgehogs" (experts who organized their analysis around one big idea and committed to confident predictions). In a follow-up "Good Judgment Project" starting in 2011, which trained ordinary people in probabilistic thinking and forecasting techniques, Tetlock found that trained forecasters outperformed the average intelligence analyst by 30% on geopolitical and economic predictions. The training curriculum included explicit attention to base rates, reference class forecasting, and willingness to update probability estimates as new information arrived -- practices that translate directly into leader decision-making contexts.

Kathleen Eisenhardt of Stanford University's Graduate School of Business studied decision-making speed and quality in a 1989 Academy of Management Journal study of eight microcomputer firms in the Silicon Valley during a period of rapid technological change. Her findings directly addressed the tension between speed and thoroughness in organizational decision-making. The fastest-deciding firms she studied were not less thorough -- they were differently thorough. They made decisions faster by using more real-time information (current operating data rather than historical analysis), consulting more advisers simultaneously rather than sequentially, building on multiple alternatives rather than narrowing to one option early, and having the CEO make fewer unilateral calls in favor of collective judgment with a designated tiebreaker. The fast-deciding firms significantly outperformed the slow-deciding firms on both financial metrics and strategic responsiveness across the study period. Eisenhardt's summary finding challenged the tradeoff assumption: better information systems and process discipline allowed speed without sacrificing quality.

Roger Martin of the Rotman School of Management, in research published in The Opposable Mind (2007) and Playing to Win (2013, co-authored with A.G. Lafley of Procter & Gamble), studied how leaders at high-performing companies framed strategic choices. His finding was that the best strategic decisions consistently arose from leaders who insisted on treating apparent dilemmas as problems to be dissolved rather than tradeoffs to be managed. Rather than asking "Should we compete on cost or quality?", the best strategic leaders asked "What would need to be true for us to compete on both?" This "integrative thinking" approach, applied systematically, produced decisions that were both more innovative and more durable than decisions made through conventional tradeoff analysis. Martin documented this pattern in case analyses of Mercedes-Benz, Four Seasons Hotels, and Research in Motion (BlackBerry), in each case showing how integrative decision framing produced competitive advantage that conventional tradeoff framing would have foreclosed.

High-Stakes Decision Cases: What the Evidence Shows

Several well-documented organizational decisions allow examination of how decision process -- not just outcome -- determined organizational performance, with consequences measurable over multiple years.

Intel's exit from DRAM memory chips in 1985-1986, analyzed extensively by Andy Grove in Only the Paranoid Survive (1996) and subsequently studied by Eisenhardt and others as a decision-process case, illustrates how organizational decision culture either enables or blocks necessary strategic choices. The decision to exit DRAM, Intel's founding business, and concentrate entirely on microprocessors required months of internal deliberation before Grove and CEO Gordon Moore reached it -- and even then, required the framing device of imagining a new CEO arriving to make the decision they could not bring themselves to make directly. The decision process was painful precisely because the existing organizational culture, which valued technical excellence in DRAM, created institutional resistance to the evidence that the business was no longer viable. Intel's market capitalization at the time of the DRAM exit was approximately $2 billion. By 2000, it exceeded $400 billion. Research by Robert Burgelman of Stanford, who had studied Intel's decision processes for years as an embedded researcher, published in the Strategic Management Journal in 1994, documented that the DRAM exit decision required bypassing several organizational mechanisms that were designed to prevent exactly this kind of rapid strategic reallocation -- a finding with implications for how organizations should design decision rights to allow necessary strategic pivots.

Ford Motor Company's decision under CEO Alan Mulally starting in 2006 demonstrates decision process design as a cultural intervention. When Mulally arrived at Ford, which was losing approximately $17 billion in 2006, he found a decision culture characterized by what he described as executives presenting optimistic reports to avoid executive punishment for bad news. His first systematic change was to a Thursday morning Business Plan Review meeting in which all senior leaders rated their performance areas on a color-coded system (green for on-plan, yellow for risk, red for problem). In his first weeks, all presentations were green -- an obvious impossibility given the company's financial position. When Mark Fields, a senior executive, showed a red indicator for a production problem, and Mulally responded with appreciation rather than punishment, the meeting structure changed. Within weeks, yellow and red indicators began appearing across the review. The behavior change enabled decisions to be made about real problems rather than managed narratives about apparent progress. By 2010, Ford was the only major US automaker that did not require a government bailout, and it earned $6.6 billion that year. Mulally's process intervention -- restructuring how information reached leadership decision-making -- was credited by multiple Ford executives as the primary driver of the cultural and financial turnaround.

The acquisition decision literature provides perhaps the most quantitatively rigorous evidence on decision quality at the organizational level. Mark Sirower of New York University's Stern School of Business, in research published as The Synergy Trap (1997) and updated in subsequent Harvard Business Review articles, analyzed 168 large acquisitions and found that approximately 70% destroyed shareholder value for the acquiring company over a three-year period. The systematic pattern of value destruction was not random -- it was predictably associated with decisions made at price premiums that required unrealistic synergy assumptions to justify. The acquiring companies' decision processes consistently underweighted integration costs, overweighted revenue synergy projections, and insufficiently stress-tested the competitive assumptions underlying the deal rationale. The implication from Sirower's data: the most consequential and irreversible decisions that organizations make -- acquisitions -- are systematically subject to the optimism bias and pre-mortem failures that the decision-making literature identifies as solvable through structured process. Companies that implemented explicit acquisition decision processes requiring documented synergy assumptions, integration plans, and independent challenge of deal rationale showed better post-acquisition performance by statistically significant margins.

References

Tags

leadership decisions decision making leadership management strategic decisions

Frequently Asked Questions

When should leaders decide alone vs. involve the team in decision-making?

The best approach depends on context: use directive (decide alone) for crises, time-critical situations, or when you have unique information; use informed (seek input before deciding) when team has expertise or buy-in is important; use collaborative (decide together) when the team has equal context or the decision deeply affects them; delegate when the decision is within someone's ownership area or they need development. Apply a framework considering impact/irreversibility, time constraints, who has information, buy-in needs, accountability, and development value. Be explicit upfront about which mode you're using to avoid confusion and frustration.

How can leaders make good decisions with incomplete information and high uncertainty?

Use the 70% rule: decide when you have roughly 70% of the information you wish you had, as waiting for 90% usually means deciding too late. Apply a structured framework: clarify the decision explicitly, identify critical unknowns, assess what you do know, gather highest-value information quickly, reason through scenarios, consider reversibility (two-way doors need less certainty), state your confidence level, and build in checkpoints for course correction. Manage cognitive biases through techniques like pre-mortems, red teams, expected value calculations, and actively seeking disconfirming evidence.

How do you build better decision-making capability as a leader over time?

Build capability through deliberate practice: keep a decision journal (record decisions, reasoning, confidence, and expected outcomes before results are known, then analyze what happened), run post-mortems on failures and pre-mortems before major decisions, actively seek diverse perspectives and dissenting views, and learn decision frameworks like first principles thinking, expected value, and reversibility analysis. Separate decision quality from outcome quality (luck matters—good decisions can have bad outcomes), study great decision-makers, and progressively tackle higher-stakes decisions. Small improvements compound significantly over a career.

What are the most common decision-making traps leaders fall into and how can you avoid them?

Common traps include: sunk cost fallacy (ask 'Would I invest if starting fresh today?'), confirmation bias (actively seek disconfirming evidence and steel-man the opposite view), groupthink (have leader speak last and assign a devil's advocate), analysis paralysis (set decision deadlines and use the 70% rule), recency bias (seek base rates not just recent events), overconfidence (calibrate with your track record and use confidence intervals), and planning fallacy (use reference class forecasting and add buffers based on historical accuracy). Use a decision checklist before major decisions to systematically check for these traps.

How should leaders communicate and explain their decisions to their teams effectively?

Communicate using the formula: Decision + Reasoning + Context = Trust. Share the specific decision clearly, explain the reasoning (problem solved, tradeoffs considered, alternatives evaluated, success metrics), provide context the team might not have (board pressure, customer feedback, strategic priorities), acknowledge the downsides and concerns honestly, and be explicit about whether the decision is final or open for input. For unpopular decisions, acknowledge the disappointment directly, explain thoroughly, and provide space for venting before moving to execution; for failures, own the decision completely and share what you learned.

Share this article