Group Decision Making Tool That Actually Works

A group decision making tool should test what the group's preferred option assumes, not aggregate which option the room prefers. Every voting matrix and ranking app on the market solves the part of group decisions that was already working and ignores the part that breaks them.

I led a group decision last year where the team reached agreement in under an hour. Nine people in the room, one option clearly preferred, no dissent worth recording. The decision failed within six months because every person in that room held the same assumption about supplier capacity and nobody surfaced it. The vote was not the failure; the failure was that the group never tested what its preferred option was resting on.

Search for a group decision making tool and every result on the page offers the same answer: voting software and ranking matrices. All of them solve the problem of choosing between options. In my experience, choosing is not the hard part. The hard part is discovering that the options share a premise the group brought into the room without examining it.

I built the Walk so that a group can separate what it knows from what it is assuming before anyone commits. The output is a Decision Document that names the critical assumption the group is carrying and the signal that would reopen the call.

A group decision making tool is a structured method that helps a team identify and test what its preferred option assumes before the group commits to it.

Why most group decision making tools miss the same thing

In 1993, Daniel Gigone and Reid Hastie gave three-member groups the task of judging student grades and measured which information actually shaped the group's verdict. Information that all three members already knew before discussion dominated the judgment; information that only one member held was systematically underweighted. They called it the Common Knowledge Effect: the number of members who shared a piece of information determined its influence on the group decision, regardless of whether that information was relevant to the task.

A voting tool amplifies this. It aggregates preferences that were already shaped by shared information; if the group entered the room with the same briefing pack, the vote reflects the briefing, not the full range of what the group actually knows. The person with a unique observation is already outnumbered before any tool asks for a preference.

NNGroup documented the same pattern in medical teams: doctors shown different video versions of the same patient spent most discussion time on what all videos showed and ignored the unique clinical observations from their specific version. The group chose the wrong diagnosis repeatedly. That is the failure mode a ranking tool cannot detect, because the ranking faithfully reflects what the group already shared.

In my experience, the pattern holds across every group setting. Project teams and executive committees spend the first half of the meeting confirming what everyone already knows and the second half choosing between options built on that shared ground.

In Deciding, Roger Estall and I list nine biases that groups carry into the room without examining. Naming a bias is useful only if the conversation then changes and the underlying assumption is tested; I have seen committees hand out bias checklists as though awareness were a countermeasure. It is not. The Walk surfaces the assumption underneath the bias and forces the group to judge whether the evidence supports it.

When a simpler group tool is enough

For a routine operational choice where the downside is containable, a voting tool or a simple ranking exercise does the job. I have run sessions where a team choosing between office layouts needed nothing more than a show of hands: the options were bounded and a wrong choice could be reversed without lasting damage.

The full method matters when a group faces a commitment where a wrong assumption reaches people or the organisation's licence to operate: a restructure that eliminates a function, or a capital investment that locks the organisation into a single supplier for years. In these decisions the group's shared blind spot carries consequences that a meeting chair cannot fix by running the session better. The group agreed and needed help checking what it had agreed on.

De Bono's Six Thinking Hats and the designated naysayer technique are useful meeting tools; they create space for dissent and offset success bias. Nevertheless, they do not establish Purpose, they do not rank assumptions by significance, and they do not define monitoring. For a group facing a routine operational decision, these techniques combined with a simple voting tool are sufficient.

Name the one assumption your group is carrying into the next decision and write the signal that would reopen it. Start the Walk →

The assumption the group already shared

In 2010, Anita Woolley and colleagues at Carnegie Mellon published research in Science showing that a group's collective intelligence is only weakly correlated with the average or maximum IQ of its members. Three factors predicted higher group performance: equal speaking turns, higher social sensitivity, and the proportion of women in the group. Teams with a dominant speaker scored lower; the collective intelligence factor accounted for 43 per cent of variance across group tasks.

A voting tool does not address conversational equality. A dominant voice can shape the options before anyone votes, and the vote then ratifies a narrowed list that already reflects that dominance. That is the structural problem a group decision making tool must solve: it must structure the conversation itself, not just the final aggregation.

That is where the trigger question in the Universal Decision-Making Method matters for groups specifically. "What are the assumptions we are making here?" asked before the group narrows to its preferred option, pulls unique knowledge into the conversation before shared ground can dominate. The Walk takes a group through that question with the conversation rules that the research says matter: open questions, inviting junior or dissenting views, and a best-idea-wins norm rather than a loudest-voice-wins default.

What a group decision making tool must actually produce

A group that votes produces a tally; a group that ranks produces a list. Neither tells the organisation what the group was assuming when it chose, or what signal should trigger a review. The Walk produces a Decision Document that names the critical assumption the group is carrying, records how the group judged the evidence, and assigns a monitoring signal and its owner before anyone commits.

That record matters for groups specifically because group decisions are harder to reopen than individual ones. An individual can reverse direction when the evidence shifts. A group that reached consensus treats the consensus itself as evidence that the decision was sound.

I have watched project teams defend a failing initiative for months because the original decision had broad support, as though the number of people who agreed made the assumption more likely to hold. The Decision Document separates the group's agreement from the group's assumptions. When context changes and a critical assumption no longer holds, the record tells whoever owns the monitoring signal to bring the decision back rather than let it run until it breaks.

I keep the Decision Document short for a specific reason: once a group record grows into a committee submission or a multi-page report, members sign the covering page and stop interrogating the assumptions underneath. A longer process protects the people who designed it, not the organisation that must live with the decision.

The Walk exists because agreement is the easiest part of a group decision to reach and the least useful part to record. What the group assumed and who will know if the assumption fails: that is what determines whether the decision survives contact with reality.

---

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