What is decision governance? Decision governance is the definition of decision rights within an organisation — who decides, who contributes input, who holds a veto, who implements, and how a decision escalates when those roles cannot resolve it.
Decision governance answers the authority question rather than the reasoning question. Its central insight is that participation and authority are different things: many people can legitimately contribute to a decision while exactly one person owns the call. Bain's RAPID model, introduced by Paul Rogers and Marcia Blenko in Harvard Business Review in 2006, separates the roles of Recommend, Agree, Perform, Input and Decide, making veto rights explicit and distinct from advisory input. GitLab assigns a Directly Responsible Individual to each decision and project, with authority and accountability held by one named person even when many contribute. Netflix designates an informed captain for significant decisions, who gathers views and expertise but ultimately decides, paired with a norm of actively farming for dissent rather than assuming silence means agreement. All three are answers to the same question. Decision governance does not assess whether the reasoning behind a decision was sound — that is decision quality — and it does not determine whether the decision was executed, which is decision effectiveness.
What Is Decision Governance?
Decision governance defines who decides, who contributes, who approves and how a decision escalates. It answers one question and one only: who has authority here?
Last updated: 2026-07-29
In short
The most expensive ambiguity in most organisations is not what to decide but who decides. Decision governance separates contribution from authority: many people can legitimately have input, while exactly one person owns the call. Everything else — meetings, escalation, approval — follows from getting that separation right.
Participation is not authority
Most decision paralysis is a governance failure wearing an analysis costume. A decision stalls not because the evidence is missing but because nobody can say, with confidence, whose call it is — so everyone consults everyone, veto rights are discovered late, and the same discussion recurs.
- •Contributing is not deciding. Being consulted does not confer a vote, and it should be said out loud at the start.
- •Veto rights must be explicit and early. A veto discovered at the end is the most expensive kind — the work is already done.
- •One decider, not a committee. Shared accountability is usually no accountability.
- •Escalation is a designed path, not an improvisation. If nobody knows how a deadlock resolves, deadlocks persist.
- •Governance is silent on quality. A clearly-owned decision can still be badly reasoned — that is a different discipline.
Three working models
These are not competing theories so much as three organisations answering the same question at different levels of formality.
RAPID — Bain
Separates Recommend, Agree, Perform, Input and Decide. Its real contribution is making Agree — the veto — explicit and distinct from advisory Input, and insisting on a single Decider. Introduced by Rogers and Blenko in Who Has the D? (HBR, 2006).
Directly Responsible Individual — GitLab
One named person holds authority and accountability for a decision or project, even when many contribute. GitLab pairs this with written decision records, because in a distributed organisation an undocumented decision effectively did not happen.
Informed captain — Netflix
For significant decisions one person gathers views and expertise, then decides. Paired with farming for dissent — actively soliciting disagreement rather than treating silence as support, which is the failure mode a single decider otherwise invites.
Where the reasoning layer fits
Governance says who decides. It is deliberately silent on what they were told — and a clear decider given a filtered picture is still going to be wrong.
Input that survives the meeting
Contributors add arguments to a structured tree rather than a thread. The decider sees the case for and against, attributed, instead of whatever surfaced in the last ten minutes.
Dissent that does not depend on courage
A question-and-answer chain makes challenging an assumption a normal move in the workflow rather than an act of confrontation — the mechanism behind farming for dissent.
A record of why, not just what
The decision, the reasons, the objections and which of them went unanswered — retrievable when the decision is reopened and the original participants have moved on.
Escalation with its context attached
When a decision moves up a level, the argument tree travels with it. The next decider inherits the reasoning rather than restarting the discussion.
Governance and reasoning are complements. Assigning an owner without improving what reaches them produces confident decisions made on poor information.
Explore further
Enterprise decision excellence
The umbrella: how governance, evidence, culture and learning combine into one system.
Decision effectiveness
Speed and execution — whether decisions get made in time and then actually happen.
Decision culture
Whether people will tell the decider something inconvenient.
Decision audit trail
How a decision and its reasoning stay retrievable months later.
RAPID vs DACI vs RACI
The decision-rights models compared side by side.
Governance & compliance
Audit-ready decision trails for regulated environments.
Frequently asked questions
What is decision governance?
The definition of decision rights in an organisation: who decides, who contributes input, who holds a veto, who implements, and how a decision escalates. It answers the authority question, distinct from whether the reasoning was sound or whether the decision was executed.
What is the difference between decision governance and decision quality?
Governance concerns authority — who decides. Decision quality concerns reasoning — whether the framing, alternatives, evidence and logic were sound. A decision can have impeccable governance and poor quality, or the reverse. They are complementary, not alternatives.
What is RAPID in decision making?
A decision-rights model from Bain separating Recommend, Agree, Perform, Input and Decide. Its distinctive contribution is making Agree — a genuine veto — explicit and separate from advisory Input, while insisting on a single Decider rather than a committee.
What is a Directly Responsible Individual?
A DRI is one named person holding authority and accountability for a decision or project, a practice documented in the GitLab handbook. Many people may contribute, but one person owns the outcome. GitLab pairs it with written decision records because of its distributed operating model.
Does consultation mean consensus?
No, and conflating them is a common and costly error. Many people can contribute views and expertise while one person owns the decision. Consensus-seeking where it was never promised produces slow decisions and resentment when someone eventually decides anyway.
How do you decide what should escalate to board level?
Typically by classifying decisions on reversibility, financial exposure, strategic importance and regulatory impact, then setting thresholds. The principle is to match the amount of process to the cost of being wrong rather than to the seniority of whoever is asking.
References & further reading
Rogers, P., & Blenko, M. (2006). Who Has the D? How Clear Decision Roles Enhance Organizational Performance. Harvard Business Review.
The RAPID model — the standard reference for separating input from authority.
View source →Blenko, M. W., Mankins, M. C., & Rogers, P. (2010). Decide & Deliver: 5 Steps to Breakthrough Performance in Your Organization. Harvard Business Review Press.
Decision rights in the context of organisational decision effectiveness.
View source →GitLab Handbook — Directly Responsible Individuals.
The DRI practice as documented by GitLab, including its relationship to written decision records.
View source →Netflix Culture — the informed captain and farming for dissent.
Netflix's own description of decision ownership paired with actively solicited disagreement.
View source →Give the decider a better picture
Clear authority only helps if the information reaching the decider is complete. Structure the arguments — including the ones nobody wanted to raise.
Start free