·Business & Strategy
Section 1
The Core Idea
Most organisations die from one of two diseases.
The first is consensus paralysis — endless meetings, watered-down compromises, decisions that satisfy no one and commit to nothing. Teams spend months debating strategy while competitors ship products. By the time everyone agrees, the market has moved.
The second is autocratic blindness — fast decisions made by one person who never heard the dissent that would have saved them. The leader decides, the team salutes, and nobody mentions the fatal flaw that three people noticed but didn't feel safe raising.
Disagree and commit is the cure for both. It separates the debate from the execution and demands excellence at each stage independently.
The principle is deceptively simple: argue your position with everything you've got. Bring data. Bring conviction. Fight for the outcome you believe is right.
Then, once the decision is made — whether or not it went your way — execute as if it were your own idea. No sandbagging. No "I told you so" waiting in the wings. No quiet sabotage through half-hearted implementation. No waiting for the project to struggle so you can say "I warned you." The commitment must be total or the model doesn't work.
Jeff Bezos crystallised this in his 2016
Amazon shareholder letter. He described greenlighting an Amazon Studios project he personally didn't believe in. Rather than block the team or demand more data, he told them: "I disagree and commit and hope it becomes the most watched thing we've ever made." The team had conviction. They'd done the work. Bezos recognised that his disagreement didn't mean his judgment was better — it meant the decision involved genuine uncertainty, and speed of execution mattered more than achieving consensus.
But Bezos didn't invent the idea.
Andy Grove built it into
Intel's culture decades earlier under a different name: "constructive confrontation." Grove expected his engineers and managers to argue with brutal directness. Hierarchy was irrelevant during debate — a junior engineer could challenge a vice president if the data supported it. The catch: once the meeting ended and a direction was chosen, everyone marched together. Grove described it as the difference between a democracy of ideas and a dictatorship of execution. Intel's 1985 pivot from memory chips to microprocessors — perhaps the most consequential strategic decision in Silicon Valley history — was born from exactly this process. Many senior leaders at Intel believed the company should stay in the memory business. Grove and
Gordon Moore decided otherwise. The dissenters committed. The company survived.
Here's the non-obvious insight most people miss: disagree and commit requires genuine disagreement first. Without real debate — without people feeling safe enough to say "this is wrong and here's why" — the model collapses into something far more toxic: comply and resent. Teams that skip the disagreement phase don't get commitment. They get performative agreement followed by passive resistance. The debate isn't a formality. It's the load-bearing structure of the entire model.
The asymmetry that makes this work is temporal. In fast-moving environments — technology, startups, any market where conditions shift quarterly — the cost of slow decisions compounds relentlessly. A wrong decision made quickly is often recoverable. A right decision made six months late is often worthless. Disagree and commit accepts a higher error rate on individual decisions in exchange for dramatically higher execution velocity across all decisions. The math works because most business decisions are reversible, and the ones that aren't are rarely the ones bogged down in consensus anyway.
Bezos drew the distinction sharply: Type 1 decisions are irreversible — one-way doors you can't walk back through. Type 2 decisions are reversible — two-way doors where you can course-correct quickly.
Most decisions are Type 2, but organisations treat them all as Type 1 — applying heavy deliberation and multi-stakeholder consensus to choices that could be tested, learned from, and adjusted in weeks. Disagree and commit is designed for Type 2 territory. It says: we've heard the arguments, we don't all agree, and that's fine. Ship it. Learn. Adjust. The alternative — waiting for unanimity on a reversible decision — costs more than being wrong.
There's one more dimension that makes this model unusual: it's recursive. The person asking others to commit might also be the person who disagreed. Bezos made this explicit. He wasn't always the decision-maker overriding dissenters — he was sometimes the dissenter choosing to commit. That symmetry matters. When a leader demonstrates that they'll commit to decisions they argued against, it signals that commitment isn't about hierarchy. It's about velocity.
What separates this from mere obedience is the quality of the disagreement that precedes it. In a healthy disagree-and-commit culture, the debate sharpens the decision. The dissenter's arguments identify risks that get mitigated during execution. The concerns that didn't change the direction still change the implementation. A marketing team that argued against a product launch timeline doesn't just comply with the deadline — they flag the specific risks they identified (insufficient channel preparation, weak messaging for a key segment) and the execution team builds contingency plans. The disagreement made the committed plan better than either the original proposal or a consensus compromise would have been.
