Discover–Discuss–Decide (Nominal Group Meeting Protocol)
Only discussion belongs in the meeting; discovery and decision happen independently and asynchronously
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 95%
Most teams do three things in a meeting — discover opinions, discuss them, and decide — and the first and third are corrupted by cross-influence from the loudest and most confident voices. Duke's fix is to collect every participant's judgments, forecasts and rationales privately before anyone talks, distribute the raw spread to the group, run a meeting that does nothing but discuss (focused on disagreement), then decide via a private vote or a single decision-maker. The group still works as a group — it just works as a 'nominal group', independently and asynchronously, so the true surface area of disagreement survives contact with the room.
Origin
Annie Duke's own consulting practice, built on the nominal-group technique from group-decision research and on Daniel Kahneman's work showing that independent judgments must be gathered before discussion to avoid anchoring. Deployed with First Round Capital and Renegade Partners.
Core principles
- 01Independent opinions first: once someone hears a senior or confident colleague's estimate, their own estimate is contaminated forever
- 02The meeting is for discussion only — not for discovering what people think, and not for making the call
- 03Focus discussion on disagreement; ~80% of a typical meeting is people restating agreement in different words
- 04Alignment is a fiction — ten different people will not leave a room agreeing, and expecting them to makes the meeting coercive
- 05Speak to convey why you believe what you believe, not to convince others you are right
How to run it
- 1
Name what the meeting will actually discuss
Before scheduling anything, decide the specific judgments you need from the room: a brainstorm of candidate features, a force-rank of an existing list, a timeline estimate, a go/no-go rating.
Pro tip Ask for a point estimate plus a lower and upper bound, not a single number — the spread is the signal.
- 2
Elicit opinions privately and asynchronously
Send the prompt to each person individually with an explicit 'don't reply all'. Require a rank or rating plus a three-to-five-sentence written rationale for why they put things in that order.
Pro tip For repeated decisions, use a Google Form / Airtable / Coda rubric that dumps into a sheet only the facilitator can see, so no one can read anyone else's answer.
Watch out If people can see each other's submissions, you have rebuilt the meeting by email and lost the whole benefit.
- 3
Circulate the full spread before the meeting
Send everyone's judgments and rationales back to the whole group to read in advance, so the meeting starts with the real distribution of opinion visible.
- 4
Facilitate discussion by reflecting back, never by agreeing
Call on each person to state their estimate and reasoning. As facilitator, never say 'I agree' — restate what you heard in your own words and ask 'is that what you meant?', letting them confirm or correct. Others may ask clarifying questions only.
Pro tip Reflect-back-and-confirm is what makes people feel heard and endowed to the decision even when it goes against them.
Watch out Interrupting silences people; 'you're wrong, here's why' makes the room tribal and people stake their ground. Both are coercive.
- 5
Decide outside the room
Make the call with a single decision-maker, or with a private vote form where no one can see how anyone else leaned. Do not decide live in the meeting.
Pro tip Above a group of one, stop expecting agreement — a partnership of six can vote independently and still not converge, and that is fine.
- 6
Use the on-the-fly version when opinions start leaking
When a new idea surfaces mid-meeting and people start blurting estimates, stop the room and have everyone write their answer on paper before anyone speaks.
In the wild
Instead of a room shouting out how many sprints a feature will take, the lead emails each person: what is your point estimate, lower bound and upper bound, and why? Everyone answers privately. The spread comes back to the group, and the meeting is spent only on where estimates diverge.
→ A junior engineer's four-week estimate actually reaches the room instead of being silently conformed to a senior's three-month number, widening the visible surface area of disagreement.
Duke ran a nominal-group discussion at a client on whether a remote policy should be consistent across functions. Opinions were collected first, then in the room she called on each person, reflected their argument back ('what I heard you say is that not all functions are created equal...'), and let them confirm or correct — offering no opinion of her own.
→ Participants felt heard and became endowed to the decision, and — seeing the true spread — accepted that nobody gets everything they want because the problem is genuinely hard.
Common mistakes
Running discovery in the room
Group brainstorming and live estimate-calling create cross-influence: the loudest and most confident person gets outsized weight, which is only fine if they happen to also be right.
Chasing alignment
Making agreement the goal turns the meeting coercive and produces a false consensus that isn't reality. People don't actually agree; pretending they do just feels better.
Talking to convince rather than to convey
The moment participants argue to win, the room becomes persuasion theatre and open-mindedness dies.
Is it for you?
Best for
Product leaders, founders and investment partnerships making repeated judgment-heavy calls (roadmaps, prioritization, estimates, hiring, deal votes) with a group of 3–15 people
Not ideal for
Fast, low-stakes, reversible calls where the coordination overhead exceeds the value, or true emergencies needing an immediate single call
From the transcript
“only thing that's ever supposed to happen in a meeting is the discussion part”
“there's cross influence the loudest person in the room tends to then have a”
“you're still working as a group but you're working as a group where you're not in the same room together and talking at the same…”
“if the goal is alignment if the goal is agreement then the meeting becomes coercive and you never want that”
“about my ideas is to convey why I believe what I do not to convince anybody that I'm right”
From the episode
This will make you a better decision-maker
Annie Duke (author of “Thinking in Bets” and “Quit,” former