A Collaborative Decision Making Model Needs One Decider

I once reviewed a project pack built around a RACI chart with 17 names and a 47-item risk register. After two committee meetings and a great deal of careful nodding, nobody could tell me who would decide whether the outsourcing should proceed if service quality slipped in month three. That is a collaborative decision making model in many organisations: a picture of participation with nobody left holding the call.

A collaborative decision making model is the structure a group uses to reach one accountable choice, from framing the decision to testing the assumptions behind it.

That is the standard the Universal Decision-Making Method was built to meet. It names one Decider and keeps monitoring inside the call, not in the appendix nobody reads.

Most models go wrong when people treat the picture as the decision. RACI can clarify contributions after the decision is framed. Delphi can collect views when status or distance would poison a live meeting. Neither tells the room what must be true for the preferred answer to work. Boxes and rounds do not decide anything.

Roger Estall and I wrote Deciding because we were tired of watching committees turn process into shelter. Consultants sell the shelter and in-house 'risk' teams keep it standing. Board secretariats then treat the shelter as ordinary furniture. Oddly enough, the people paid to preserve it never recommend leaving it (a heartbreaking coincidence, I am sure). I wanted a method that left one person exposed to reality and forced the room to say what the decision was betting on.

Most collaborative decision making models start in the wrong place

Click to expand

Most collaborative decision making models begin with machinery, and that flatters the people who live off the machinery. McKinsey's critique of RACI noted in 2022 that 80 per cent of organisations struggle with decision making and that bloated role charts slow the call instead of sharpening it. I do not need the survey to believe it, but the survey is correct. By the time a room has argued over the boxes, it may still not have stated the choice or the outcome it wants. The life of the decision can be settled afterwards.

Model	Useful for	Who owns the call	What gets missed
RACI	Clarifying contribution	Often blurred	What decision is actually being made
Delphi	Gathering dispersed expert judgment	Usually deferred	When there is enough certainty to act
Universal Decision-Making Method	Running the decision itself	One Decider	The room contributes, but one person still owns the call

I do not object to RACI as an aid to implementation. I object when it becomes the frame. If the chart arrives before the decision has been properly framed, administration has already beaten judgment.

A collaborative decision making model needs one Decider

The evidence for collaboration is useful, but it is not mystical. Rutka, Wrobel, and Wycinka looked across 598 teams and found that consultation and group discussion reduced error against autocratic calls, yet neither reliably beat the best informed individual in the room. The room can improve the input. It does not absorb accountability.

Clients have said to me, "Grant Purdy, you do bang on about one Decider." I do, because the Decider is not automatically the most senior person. It is the person who has the authority to act and who cannot disappear when the outcome turns sour. Committees dislike that arrangement because it denies them the usual refuge of "we all agreed", which is often how nobody agrees to own anything.

In my experience, "shared accountability" is a laundering device. It suits consultants and governance teams because it spreads exposure so thinly that the paperwork looks prudent while the decision remains homeless (though the minutes look splendid). A committee can congratulate itself on the quality of the discussion and still leave the hard exposure on nobody's desk. That is why effective decision making in teams depends on extracting challenge for one accountable call rather than preserving harmony.

RACI and Delphi are tools, not full decision models

RACI can help after the frame is set, and Delphi can help when status or geography would spoil a live meeting. I still would not call either a full decision model. Delphi is particularly attractive to institutions that want disciplined input without the inconvenience of a visible owner. It can pool views. It cannot tell me who owns the call or which signal should bring the decision back for review.

The reporting quality tells the real story. Schifano and Niederberger reviewed 287 Delphi studies in the health sciences and found that 61 per cent did not define consensus before the work began, while about half did not report a clear stopping rule. That kind of reporting is procedural camouflage dressed up as rigor. If nobody can say what counts as enough, the panel can keep polishing its agreement until the calendar finally makes the decision for them.

That is one reason groupthink in decision making survives inside respectable organisations. Once convergence becomes the goal, disagreement gets treated as bad meeting behaviour instead of evidence that an assumption is still loose.

The collaborative decision making model I trust tests assumptions

This is the step the fashionable models usually skip. Most groups move from options to preference without naming what the preferred answer assumes. I want the room to ask two blunt questions: what has to be true for this to work, and how much pain do we buy if it is false? Once those assumptions are on the table, the conversation gets less theatrical very quickly. People can no longer hide behind adjectives or percentages. They have to defend the bet itself.

When I run this in a room, I do not start with "thoughts?" I start with the decision itself. What exactly is being decided, who owns it, and what would count as a good outcome? Once that is plain, the room stops drifting into commentary. One person owns the call. Everyone else is there to improve it. I want the people closest to the work saying what this answer is assuming, and I want the awkward assumption said in ordinary English before somebody hides it under a chart.

The Decider then has a real job. They judge when the important assumptions have been pushed hard enough that acting is safer than waiting for another perfect memo that will never arrive. Anyone selling perfect certainty is back in the furniture business. Before anyone leaves, the room settles what will be watched, who will notice if the context shifts, and what kind of change opens the decision again. If that part is left for later, it usually means never.

I once watched a plant upgrade approved on the strength of a shutdown date buried in an appendix. The room treated the outage window as an administrative entry, not as an assumption that needed testing against contractor readiness and the state of the site preparation. Nobody asked what had to be true for that date to hold. When the window moved, the economics that justified the project collapsed with it. That was a failure to drag one critical assumption into daylight before the decision stood.

A collaborative decision making model worthy of the name does not ask whether everyone had a turn. It asks whether one accountable Decider has enough certainty to act because the room has exposed the live assumptions and settled what to watch next. That is the version of collaborative decision making I trust. The rest is committee theatre, useful mainly to the people paid to keep the theatre open.

---

Grant Purdy is the co-author, with Roger Estall, of Deciding (2020), and the architect of the Universal Decision-Making Method.