RAPID: Decision Rights

Six of you are standing on the pavement deciding where to eat.

Everyone has an opinion and everyone offers it generously. Somebody mentions the place with the good noodles, somebody else is not in the mood for noodles, somebody is checking whether the other place is open, and two people are saying they genuinely do not mind while continuing to raise objections. Twenty minutes go by. Nobody has been unreasonable and nobody has disagreed with anybody, and you have not moved.

Then one person says: I am picking. Any hard nos? Two seconds of silence, and you walk.

Nothing changed about the information. What changed is that one person took the decision and everyone else’s contribution became input rather than a vote. Ambiguity about who decides is more expensive than deciding wrong, and it is worse than deciding wrong because it is silent: there is no meeting where anyone says “we failed to decide”, and everybody involved feels they behaved well.

The five roles#

Bain’s answer is RAPID, five roles assigned per decision. The letters are not in process order, which is a known irritation of the model and worth naming so nobody tries to read it as a sequence.

Two of these carry almost all the value, and they are the two people usually skip.

D is a single person. Not a committee, not a team name. If two people can each believe they hold D, you have the pavement.

I is not a veto. This is the distinction that makes the model worth adopting. Being consulted means your view must be genuinely sought and considered, and it does not mean the decision waits for your agreement. Teams that quietly treat input as a veto end up with decisions that require unanimity, which is how a group of reasonable people produces the slowest possible organisation.

A is deliberately rare. Agree is for people who can legitimately stop a decision, typically because of a legal, regulatory or contractual constraint. Handing out A generously is the most common way this model gets ruined, because every additional A is another place the decision can stall indefinitely.

Assigning it takes one line#

The overhead people fear is imaginary. For most decisions this is a single line at the top of a document:

R Sari · D Andro · A none · I engineering, support · P the payments team

Written before the analysis, not after. Assigning decision rights after a conclusion has appeared is how the role assignment becomes an argument about the conclusion.

How to tell#

Ask two people in the room who decides. If you get two different answers, you have found the real problem and it is not the one on the agenda. This takes ten seconds and is the single highest-yield check on this page.

Count the As. More than one or two and the decision cannot move. For each one, ask what specifically they are entitled to block and on what grounds. If the answer is “they would be upset”, that is I, not A.

Watch what happens to the person who was overruled. If someone’s input was heard and not taken, and the decision then quietly stalls or gets relitigated next week, input is functioning as a veto regardless of what the document says.

Notice decisions that keep coming back. A decision revisited three times was never made. Either D was unclear, or D made a call and did not communicate it as one.

When it goes wrong: the most common failure after adopting RAPID is treating it as a process to be run rather than a line to be written, which turns a thirty-second clarification into a workflow and gets it abandoned within a month. If assigning it takes longer than the decision deserves, do not assign it.

Where this stops#

It is overhead on reversible decisions. Most decisions should be made by whoever notices them, immediately, without ceremony. Reserve this for decisions that are hard to reverse, cross more than one team, or have already stalled once. Applying it to everything is worse than not having it, because it teaches people that deciding requires paperwork.

It does not make a bad decider good. RAPID allocates authority. It has no view on judgement, and a clear decision right in the hands of someone with poor judgement produces faster bad decisions.

It says nothing about how the decision is made. The framing, the options, and the evidence are the work of issue trees and the two-pager. RAPID only settles who.

Disagree and commit still has to be taught separately. The model gives someone the right to overrule; it does not give the overruled person a way to back a decision they lost. That is a cultural norm, not a role assignment, and it is the part that actually makes single-decider systems bearable.

How we use it here#

One line at the top of any two-pager or decision document, written before the analysis. For ongoing ownership rather than a one-off decision, DRI is the standing version of the same idea, and the two should agree: the DRI for an area is usually D for decisions inside it.

The check to run in meetings is the ten-second one. When a discussion has gone round twice, stop and ask who decides this. Most of the time the answer exists and simply has not been said out loud, and saying it ends the discussion. When the answer does not exist, that is the thing to fix, and it is a better use of the remaining time than continuing to discuss the substance.

Related#

• DRI is the standing version: who owns an area, not a single decision.
• Two-pagers is where the RAPID line goes.
• Structured Problem Solving covers how the decision gets framed.
• McKinsey 7S treats unclear ownership as a structure problem when a change will not stick.