Disagree and Commit, Properly: What It Asks of Both Sides

“Fine.”

Picture a pricing meeting. The commercial director proposes ending the standing 12% discount for mid-size accounts. The head of sales objects: two large customers will walk, he says, and his team will spend a quarter firefighting. The debate runs forty minutes. The CEO sides with the commercial director. The head of sales says, “Fine.”

Six weeks later the discount is still being quoted. Not everywhere. Just in the deals where the reps judged it necessary, which turns out to be most of them. Nobody broke a rule. Nobody reopened the decision. It simply did not happen.

Disagree and commit is the practice of arguing your view fully before a decision, then, once it is made, putting your full effort behind it as if it had been your idea. It is usually told as a rule for the person who lost the argument. That is half of it.

The thesis of this page: disagree and commit is a deal with two sides. The dissenter owes full effort, out loud. The decider owes a real hearing, a stated reason, and a date and a test for looking again. Run one-sided, as a gavel, it produces what happened in the pricing meeting: compliance in the room and quiet non-execution outside it.

Commitment is bought with a hearing.

Where the phrase comes from

The best-documented source is Amazon. One of its leadership principles is titled “Have Backbone; Disagree and Commit.” It says leaders “are obligated to respectfully challenge decisions when they disagree, even when doing so is uncomfortable or exhausting,” that they “do not compromise for the sake of social cohesion,” and that “once a decision is determined, they commit wholly” ².

Jeff Bezos explained the phrase in his 2016 letter to shareholders. When someone has conviction about a direction without consensus, he suggests asking: “Look, I know we disagree on this but will you gamble with me on it? Disagree and commit?” He adds, “This phrase will save a lot of time,” and, notably, “If you're the boss, you should do this too” ¹.

Netflix describes a version of the same idea in its culture memo. An “informed captain” is expected to seek out different opinions at every level, which it calls “farming for dissent,” and then everyone, including the people who argued for a different approach, is expected to “disagree then commit” ³.

For how disagree and commit fits into a whole decision process, including when to escalate instead, see getting a room to decide. This page goes deeper on the commitment itself: what makes it real, and why it so often is not.

The two sides of the deal

The right-hand column is not etiquette. It is what makes the left-hand column possible. W. Chan Kim and Renée Mauborgne, who wrote about fair process in Harvard Business Review ⁴, describe three principles: engagement, involving people in the decisions that affect them by asking for their input and letting them challenge each other's ideas; explanation, making sure everyone involved understands why the final decision was made; and expectation clarity, stating the new rules of the game ⁵. Their argument is that people who trust the process was fair cooperate voluntarily in execution, even when their own ideas were rejected ⁵.

Read the decider's column as fair process applied to a single meeting. The hearing is engagement. The stated reason is explanation. The assignments and the review date are expectation clarity.

Why “fine” is not commitment

The head of sales in the opening did not lie. He said “fine” and meant “I heard you.” He did not mean “I will now spend my credibility with my team selling a change I think is a mistake.” Those are different promises, and the meeting never asked for the second one.

Three mechanisms turn a lost argument into quiet non-execution.

• The dissenter has a stake in being right. If the decision fails, the person who opposed it is vindicated. Nobody needs to sabotage anything for that incentive to work; it only takes a little less energy behind the change than it needed.
• Commitment that is not stated is not owned. A nod in a meeting creates no obligation anyone can point to later. A sentence spoken by name (“I disagree, I've said why, and I'm committing to make this work”) does.
• Decisions without a review date never close. If there is no agreed moment to look again, the only way to keep an objection alive is to keep acting on it. A review date gives the objection somewhere legitimate to wait.

Bezos's description of his own commitment on an Amazon Studios project shows the standard. He wrote that it was not a case of privately concluding the team was wrong but not worth chasing. It was “a genuine disagreement of opinion, a candid expression of my view, a chance for the team to weigh my view, and a quick, sincere commitment to go their way” ¹. Every element is visible. Nothing is left implied.

Disagree out loud before. Commit out loud after.

How to run it in the room

1. Ask for the strongest objection before decidingName the people most likely to disagree and ask them directly. Silence in a meeting is not agreement; it is usually a calculation about whether speaking up is worth it.
2. Restate the objection until its owner agreesThe decider, or the chair, says the objection back in their own words. When the dissenter says “yes, that's it,” the hearing has happened, and everyone saw it.
3. Decide and give the reasonSay what was decided and why the objection did not change it. “We weighed the risk of losing two accounts against the margin across 400, and the margin wins” is a reason. “We've discussed it enough” is not.
4. Ask for commitment by nameGo around the people who must act. Each says whether they commit. If someone cannot, find out now whether it is a disagreement (commit anyway) or a line they cannot cross (escalate or address it).
5. Set the review date and the testAgree when you will look again and what evidence would reopen the decision. In the pricing case: lost revenue from the two named accounts above a set amount by the end of the quarter.
6. Write it downOne paragraph: the decision, the reason, the dissent and who held it, the review date and the test. A decision log entry is enough.

The pricing meeting, run again

Look at what the head of sales got in exchange for his commitment: his estimate on paper, a test that could prove him right, and a date. He did not need to keep the objection alive through his team's behavior, because the process was keeping it alive for him.

When not to commit

Disagree and commit is a tool for differences of judgment. It is the wrong tool for three other situations.

• Ethics and law. If you believe a decision is illegal, unsafe or dishonest, the obligation is to escalate or refuse, not to commit. No version of the principle asks otherwise.
• Deep misalignment. When two parties have different objectives, Bezos writes that no amount of discussion will resolve it, and without escalation “the default dispute resolution mechanism for this scenario is exhaustion. Whoever has more stamina carries the decision” ¹. His advice is to recognize true misalignment early and escalate it immediately ¹. Asking someone to commit to an objective they do not share is asking them to fail politely.
• New information. Commitment is to a decision made on the evidence available. When material new evidence arrives, raising it is part of the deal, not a breach of it. The test agreed at the time is what separates new information from a rerun of the old argument.

Now the strongest objection to the whole practice: disagree and commit is a tool the powerful use to end arguments they are losing. The boss says “disagree and commit,” the debate stops, and the subordinate's dissent is converted into a public pledge of loyalty. Critics who have seen it used this way are not wrong.

The answer is that the misuse is detectable, because it is one-sided. Three tests tell you which version you are in. Does the boss ever disagree and commit to a subordinate's call, the way Bezos says the boss should ¹? Does the dissent get written down with its owner's name, so it can be vindicated? Is there a review date? If the answer to all three is no, the phrase is being used as a gavel, and the commitment it produces will be the kind that says “fine.”

Making it a habit across a team

One meeting can be run well by a good chair. A culture of disagree and commit needs the senior team to model the right-hand column of the table visibly, and especially the hardest part: the leader committing to a call they argued against. Amazon's own principle frames the obligation to challenge as applying to leaders ², and Bezos's letter tells the boss to do it too ¹. When a team sees its leader lose an argument and then champion the result, it learns the deal is real.

Two second-order problems show up at scale. First, dissent dries up when it is costly. If people who objected to failed decisions are not credited when the review vindicates them, the next objection will not be voiced. Second, “commit” can harden into “never revisit.” Teams that review on the agreed date and actually reverse some decisions keep the practice honest; teams that never reverse anything teach people that the review date is theater.

If decisions keep getting relitigated, the problem is usually upstream: unclear decision rights. RACI, RAPID and DACI each settle who decides in a different way, and a team that knows who holds the decision has less to fight about afterward. For the team dynamics behind avoided conflict and weak commitment, the Five Dysfunctions model is a useful lens.

Monday morning: pick the last decision your team made over an objection. Ask the person who objected, privately, whether they have committed or complied. If you do not like the answer, the fix is not to ask for more loyalty. It is to give them what the deal promised: a hearing they recognize, a reason, and a date when the evidence gets its turn.

Commitment cannot be ordered. It is traded for a hearing and a fair chance to be proven right. Make that trade explicit and “fine” becomes a promise someone intends to keep.

Questions

What does disagree and commit mean?

It means arguing your view fully before a decision and then, once the decision is made, committing your full effort to making it work, even though you argued against it. Amazon lists it as a leadership principle, and Jeff Bezos's 2016 shareholder letter describes it as a way to move quickly when there is conviction but no consensus, adding that the boss should do it too.

Who came up with disagree and commit?

The phrase is widely credited to Intel under Andy Grove, but we have not found it in a primary Intel or Grove source, so that attribution is unconfirmed. The best-documented sources are Amazon's leadership principle “Have Backbone; Disagree and Commit” and Jeff Bezos's 2016 letter to shareholders. Netflix's culture memo uses a similar phrase, “disagree then commit.”

When should you not disagree and commit?

When the issue is ethical or legal, when the parties have genuinely different objectives that discussion cannot reconcile (escalate early instead), and when material new information arrives after the decision. Disagree and commit is for differences of judgment about the best way forward, not for disputes over whether a decision is right to make at all.

How do you make disagree and commit work?

Treat it as a two-sided deal. Ask for the strongest objection before deciding and restate it accurately. Give the reason for the decision. Ask each person who must act to commit out loud. Agree a review date and the evidence that would reopen the decision, and write down who dissented so they can be credited if the evidence goes their way.

Is disagree and commit just a way for bosses to shut down debate?

It can be used that way. The misuse is one-sided: dissent is silenced but never written down, there is no review date, and the boss never commits to anyone else's call. When the decider also gives a real hearing, a reason and a date to look again, and occasionally commits against their own view, the practice protects dissent rather than ending it.

Where to go next

• Getting a Room to Decide: A Decision-Making Framework
• RACI vs. RAPID vs. DACI: Which Decision Model, When
• Decision Log
• Lencioni's Five Dysfunctions
• Radical Candor