You have been in the room: a recurring two-hour workshop, a dozen people, a screen full of sticky notes, and at the end a vague "let's take this offline" that decides nothing. The fix is rarely a better icebreaker or a slicker template. A session works when it is designed, when someone decided its purpose, planned how the group would move from many ideas to one decision, and held the room through the uncomfortable middle where that movement actually happens.

The quick version

  • Purpose before agenda. Decide what the session is for, specifically, before you pick activities. "A category is not a purpose", a planning meeting is not a purpose.
  • Design the group's journey, not just the slides. Good sessions open up (divergence), pass through a hard middle, then narrow to a decision (convergence). Skipping the middle is why decisions don't stick.
  • The "groan zone" is the work, not a failure. The frustrating part where ideas collide is where shared understanding is built, your job is to hold the group in it, not rescue them out of it.
  • Structure beats charisma. Amazon replaced presentations with memos read in silence because format, not personality, drives the quality of the discussion.

The idea in depth: purpose is the design, not the preamble

The most useful starting move in meeting design comes from Priya Parker, a conflict facilitator, in The Art of Gathering: How We Meet and Why It Matters (2018). Parker's core claim is blunt: most gatherings are mediocre because the host skips the question of why and jumps straight to logistics. Her line, drawn from interviews with more than a hundred gatherers, from Quaker clerks to circus choreographers, is that "a category is not a purpose." "We need a planning meeting" names a category; it tells you nothing about what a good outcome looks like. She pushes hosts to keep asking why until the real purpose surfaces, then let that purpose decide everything else: who is in the room, how long it runs, what happens in it.

In practice that means writing the session's purpose as a sentence of consequence before you book a room or open a slide. Not "Q3 roadmap workshop" but "by the end, we will have agreed which two initiatives we're cutting to fund the migration." That sentence is a test you can run every activity against, if a segment of the agenda doesn't move the group toward it, cut it. This is the same answer-first discipline that runs through structured communication: decide the outcome, then build backwards to it.

An honest limitation. Parker is a practitioner writing from craft and observation, not a controlled study, her book is a well-argued field guide, not peer-reviewed evidence. Treat "purpose first" as a reliable heuristic, not a proven law. It is, however, the one move almost no bad meeting makes.

How groups actually move: divergence, the groan zone, convergence

Purpose tells you where the room should end up. It doesn't tell you how a group of people gets there, and groups don't move in a straight line. The clearest map of the journey is the Diamond of Participatory Decision-Making from Sam Kaner's Facilitator's Guide to Participatory Decision-Making (3rd ed., 2014). Kaner describes three zones. First the divergent zone, where the group generates options and surfaces different perspectives. Then a groan zone in the middle, the uncomfortable stretch where those perspectives grind against each other, people repeat themselves, and the room feels stuck. Then, only after that, the convergent zone, where the group narrows toward a workable decision.

flowchart LR
  A(["Frame:
purpose + question"]) --> B(["Diverge
generate options, hear all views"])
  B --> C(["Groan zone
views collide, feels stuck"])
  C --> D(["Converge
narrow toward a workable option"])
  D --> E(["Close
decision + owner + next step"])
Kaner's diamond: a group has to widen before it can narrow, and the hard middle is part of the route. Leaders Loop

The practical payoff is that the groan zone is not a sign the meeting is going badly, it is the meeting working. Kaner's insight is that an idealised, tidy process tries to skip it, and that's exactly why those decisions don't hold: the group never built shared understanding, so people leave with private reservations they voice later. So the move is to name the groan zone when it arrives ("this is the messy bit, it's normal, stay with it"), resist the urge to vote early to escape the discomfort, and only push toward convergence once the real disagreements are on the table. A facilitator who collapses the diamond into a quick brainstorm-then-vote gets speed and loses commitment.

An honest limitation. The diamond is a model of how participatory decisions tend to go, not a guarantee that sitting in the groan zone always produces a better answer. Some meetings genuinely don't need it, a status update or an irreversible call the leader already owns shouldn't be dressed up as a workshop. Match the format to the decision, a point that connects to managing up, down and across: not every question is a group's to make.

Why format beats personality: the case for silence

The cliché is that great facilitators are great performers, quick, warm, magnetic. The more durable lever is structure. The sharpest example is Amazon's decision to ban PowerPoint from its leadership meetings. In his 2018 letter to shareholders, Jeff Bezos described the practice in his own words: instead of slides, teams "write narratively structured six-page memos" and "silently read one at the beginning of each meeting in a kind of 'study hall.'" Only after everyone has read does discussion begin. The reasoning is that a slide deck lets a confident presenter gloss over weak logic, while a written narrative forces the author to think the argument all the way through, and silent reading means the room debates the same, fully-absorbed content rather than reacting to whoever talks first or loudest.

"We silently read one at the beginning of each meeting in a kind of 'study hall.'", Jeff Bezos, 2018 shareholder letter

Even if you never write a six-pager, the lesson holds: engineer the room so good thinking can't be drowned out by fast talking. Send the pre-read in advance, or, more reliably, build five minutes of silent reading into the meeting itself, because the document everyone "should have read" is the document half the room didn't. Ask people to write their position before they speak it; collect ideas on cards before opening the floor. These are small structural choices, and they do more for the quality of a discussion than any amount of facilitator charm.

There's evidence the problem is real and largely invisible to the people in charge. Organisational psychologist Steven Rogelberg, in The Surprising Science of Meetings (Oxford University Press, 2019) and his related Harvard Business Review work, documents a persistent gap: leaders rate the meetings they run far more highly than the people sitting in them do. The implication is uncomfortable. If you think your meetings are fine, you are exactly the person whose meetings most need redesigning, and the cheapest correction is to ask attendees, honestly and anonymously, what they'd change.

A worked example

Take a head of operations, call her Dana, running a quarterly planning workshop with twelve people. (Illustrative example; not a real company.) Her usual version is a marathon: a full agenda, back-to-back presentations, a frantic brainstorm in the last half hour, and a list of "actions" that nobody owns. People dread it, and the same arguments resurface next quarter because they were never actually resolved.

This time she starts with purpose, Parker-style. One sentence: "By the end, we'll have agreed the three priorities for next quarter and what we're explicitly not doing." That line shrinks the guest list (the people who own the work, not the whole department) and cuts two presentations that didn't serve it.

Then she designs the journey using Kaner's diamond rather than a stack of updates.

flowchart TD
  A(["Purpose written:
agree 3 priorities + what we drop"]) --> B(["5 min silent read
of the context memo"])
  B --> C(["Diverge: each person
writes candidate priorities first"])
  C --> D{"Groan zone
named out loud"}
  D -->|"hold the room,
don't vote early"| E(["Surface the real
trade-offs & objections"])
  E --> F(["Converge:
dot-vote to 3 + a 'not doing' list"])
  F --> G(["Close: owner +
first step per priority"])
From a presentation marathon to a designed decision, purpose frames it, the diamond shapes it. Leaders Loop

She opens with five minutes of silent reading of a one-page context memo, so everyone starts from the same facts. Diverge: each person writes their candidate priorities privately before anyone speaks, so the loudest voice doesn't anchor the room. When the trade-offs collide, engineering wants stability, sales wants features, she names the groan zone instead of papering over it: "this tension is the actual decision; let's stay in it." Only once the real objections are aired does she move to convergence: a quick dot-vote narrows to three priorities, and she insists on the harder list of what they're dropping to make room. Each priority leaves with an owner and a first step. The session runs ninety minutes, not three hours, and the decisions survive contact with the next week, because the disagreement happened in the room, not in the corridors afterwards.

Frequently asked questions

What's the difference between facilitating and chairing a meeting?

A chair owns the outcome and often has a stake in it; a facilitator owns the process and stays neutral on the content. The distinction matters because the moment you're seen steering toward your own preferred answer, people stop contributing honestly. If you genuinely need to be both, common for a team lead, say so explicitly, or hand facilitation to someone else for the contentious parts so the group trusts the process is fair.

How do I stop one or two people dominating the room?

Design for it rather than policing it in the moment. Ask people to write before they speak, so quiet thinkers have a contribution ready and the fast talkers can't anchor the discussion. Go around the table deliberately, use silent-reading and silent idea-generation, or collect input on cards first. Structure does the work that repeatedly interrupting a dominant voice can't, and it does it without anyone feeling told off.

Do we really need a workshop, or would an email do?

Most things that get booked as meetings shouldn't be. A useful test: a live session earns its cost when the group needs to think together, to surface disagreement, build shared understanding, or make a decision people must commit to. If you only need to broadcast information or collect individual input, a document or an async thread is faster and respects people's time more. Bring people together for the groan zone, not for status updates.

How long should a workshop be?

Let the purpose and the decision set the length, not the calendar's default 60-minute block. Attention degrades fast, so a tightly designed 90 minutes usually beats a sprawling half-day. If you genuinely need longer, break it into distinct phases with real breaks, and be honest about which phase each segment serves, divergence, the hard middle, or convergence. A long meeting with no shape is just a short meeting that wouldn't end.

Isn't all this structure just over-engineering a conversation?

It can be, if you apply a full diamond to a five-minute decision. The skill is matching the design to the stakes: a quick check-in needs almost none, a genuinely contested cross-functional call needs a lot. The structure isn't there to make a simple thing complicated; it's there to stop a complicated thing collapsing into the loudest opinion or a decision nobody actually agreed to.

Related in the Toolkit

Facilitation is where several communication skills meet a live room, it leans on the answer-first habit of structured communication and the framing instinct of managing up, down and across, then asks you to hold a group through disagreement.

Where to go next