Disagree and Commit
Argue openly until the decision is made, then everyone backs it fully — including those who argued against it.
“I disagree and commit and hope it becomes the most watched thing we’ve ever made.”
To an executive
Decisions stall when everyone must agree. A fair hearing, a clear decider and full support afterwards lets the organisation move faster without losing the benefit of dissent.
To an engineer
Raise objections with evidence before the decision is made. Once it is made, build it properly, and reopen it only when new facts appear.
The idea
The phrase is associated with Intel under Andy Grove and is one of Amazon's leadership principles. Jeff Bezos explained it in his 2016 letter to shareholders: when you are confident a direction is right but not everyone agrees, ask people to disagree and commit, and be willing to do the same yourself.
It has two halves, and both matter:
- Disagree — challenge the proposal openly and with evidence while the decision is still open.
- Commit — once it is made, support it fully. No quiet foot-dragging, no "I told you so" in the corridor.
Why it works
It separates the debate from the delivery. Teams get the benefit of dissent before the decision and of alignment after it.
It stops endless relitigating. Many architecture decisions are good enough either way; what kills them is reopening them every sprint.
It lets decisions move faster. You do not need consensus, only a fair hearing and a clear decider.
Where it applies
- Architecture decisions with several viable options, where waiting for unanimity costs more than either choice.
- Cross-team standards, where some teams would have chosen differently.
- Leadership forums, where the architect may be the one asked to commit.
How to apply
- Make the decider and the deadline clear before the debate starts.
- Invite the strongest objections, and make sure they are heard and recorded.
- Decide, and say so explicitly, with the reasons — an ADR is a good place.
- Name the conditions for revisiting it, so dissenters know what evidence would reopen the question.
- Commit in public, especially if you argued against it.
Example scenario
Two senior engineers disagreed on whether the new platform should standardise on gRPC or REST. After a week of debate, the principal architect chose REST for external APIs and recorded the decision, the gRPC case, and a trigger: revisit if internal latency targets were missed.
The gRPC advocate wrote the team's REST guidelines. Six months later, latency was within target, and the question never came back.
Questions to ask
- Has everyone had a real chance to argue their position?
- Who makes this decision, and when?
- What would make us revisit it — and is that written down?
When not to use it
- Ethical, legal or safety concerns. Commitment is not a reason to stay silent about something harmful; escalate instead.
- One-way doors made without a real debate. If dissent was never heard, asking for commitment is just compliance.
How it misleads you
Leaders can use "disagree and commit" to skip the disagreeing. If objections are cut short or ignored, people learn to stay quiet, and the decision loses the very challenge that made it worth trusting.
In the wild
Further reading
Last reviewed · Suggest an improvement