The Anatomy of a Workshop

01 Title

On screen "The Anatomy of a Workshop" · Module 3 · Facilitation Foundations

A few years ago I watched a chef break down a whole chicken in about forty seconds. No sawing, no hacking at it. He just kept finding these places where the bird came apart, and then it was in eight pieces on the board and he was wiping his hands.

So I asked him how he got that fast. And he said something I've thought about ever since.

'I'm not cutting through anything. The bird's already in pieces. I'm just finding where they meet.'

That's this module. Every workshop you'll ever run is already in pieces. Five of them. Same five whether you've got forty-five minutes with your own team or three days offsite with a client.

Hold onto him. Hold onto the joints in particular, because we come back to them at the end.

And once you can see where they meet, you stop hacking.

02 A workshop is not a list of activities

On screen a scattered pile of activity names, no structure

Most people plan a workshop by listing activities they liked somewhere else. Crazy 8s, because it went well last time. An icebreaker, because you're supposed to. A vote at the end, because you need a decision.

And you get a day that's busy, and well-liked in the room, and produces nothing anyone can act on. I've run that day. More than once.

The problem isn't the activities. The activities are fine. Swap Crazy 8s for brainwriting and almost nothing changes.

The problem is there's nothing underneath them, and the thing underneath is what does the work.

03 The five blocks, and how long each one gets

On screen FRAME · DIVERGE · CONVERGE · COMMIT · CLOSE, with a proportions strip beneath

Five blocks. Frame, diverge, converge, commit, close.

Length changes. The blocks do not. Two hours or two days, same five, in that order.

And I want to be clear about how strong a claim I'm making, because it sounds like a framework and frameworks are cheap. Every functioning session you have ever been in had all five, whether or not anyone named them. Every broken one was missing at least one. That's the whole diagnostic.

Rough proportions while you've got the five in front of you, and I'll flag these honestly: they're mine, from practice, not from a study. In a ninety-minute session, frame about ten minutes, diverge about twenty-five, converge about thirty, commit about fifteen, close about ten.

Two things worth noticing. Converge usually needs more time than diverge, and almost everybody plans it the other way round because generating is more fun. And commit plus close is a quarter of your session, which will feel like far too much until the first time you protect it and watch things actually happen afterwards.

04 Frame

On screen FRAME · "Why we're here, what we'll produce, how we'll behave"

Frame. Why we're here, what we're going to produce, and how we're going to behave. Two minutes of it or twenty, depending on the room and how well they know each other.

The frame is where you spend the trust you're going to need later, and it's the block people cut first because it feels like preamble.

Concretely it's: the objective read out loud, the shape of the day, the roles, the working agreements, and what happens to the output afterwards. That last one matters more than people think. A room that doesn't know where the work is going invests less in making it good.

0:42 · then: "You spend all day re-earning permission you could have had once."

And skipping it fails in a way that's genuinely hard to spot, because it never shows up as a single moment.

Every instruction you give afterwards gets a little bit questioned. Why are we doing this bit? How long is this going on for? Nobody's being difficult. They just never got told the rules, so they're working them out live, one activity at a time.

You spend the whole session re-earning permission you could have had once, at the start, for free. And you'll experience that as the room being resistant, which is the wrong diagnosis, and it'll make you push harder, which makes it worse.

05 Diverge, and the mistake hiding inside it

On screen DIVERGE · "Widen the field" · "Discussion is not divergence."

Diverge. Widen the field. More options, more perspectives, more of the problem visible than anybody carried in through the door.

The test for a real divergence is that the room ends up holding something nobody brought in with them. If everything on the wall was already in somebody's head at nine o'clock, you didn't diverge, you collected.

Skip it and the group picks between the two ideas the loudest person already had before the meeting started. Which they could have done by email, in about four minutes, without booking a room and a projector.

There's also a specific way this block goes wrong that catches nearly everyone, and it's worth naming.

Discussion is not divergence.

An open discussion feels like generating. It isn't, quite. In an open discussion, the ideas that get into the room are the ideas belonging to people comfortable speaking in that room, and everybody else's stay in their heads. You get the same three voices' option set, more slowly.

Real divergence is almost always written and almost always silent to start with. That's module six in full, but the design implication is here: when you write 'diverge' in your agenda, write a structure next to it, not the word 'discussion.'

06 Converge

On screen CONVERGE · "Cluster, compare, cut, choose"

Converge. Cluster, compare, cut, choose.

This is the block people are most afraid of, and the one they're most likely to fake, because it's the block where somebody's idea doesn't make it. That's genuinely uncomfortable, and there's a strong pull to avoid the discomfort by never quite choosing.

Faking it looks like a vote with no discussion first. Or a facilitator summarising what they think the group meant, which is module one's failure wearing a different hat. Or 'let's take all of these forward', which is a decision not to decide.

0:36 · then: a wall of stickies, a phone camera

Skip converge properly and you know exactly what you leave with, because you've seen it.

A wall of sticky notes and a photograph.

Somebody takes it. Somebody posts it in a Slack channel with 'great session today!' And nobody opens it again, ever, including the person who took it.

I've got a folder of those photographs. Six years of walls I was going to synthesise later. Full disclosure, I've never once gone back to one of them and turned it into anything.

07 Commit, and why it's the block that goes missing

On screen COMMIT · "Who does what, by when" · "You're always out of time at the end."

Commit. Turn the choice into ownership. Who's doing what, by when, said out loud, in front of everybody, with a name attached.

And every word in that sentence is load-bearing. Not 'we should look into it.' Not 'the team will pick this up.' A person, a thing, a date, spoken aloud while the rest of the room is listening.

Out loud matters because a commitment made in front of eight people behaves completely differently from the same commitment typed into a doc afterwards. You're not creating a task. You're creating a small social fact. Without this block, the whole thing was a very expensive conversation.

Now think about the last long meeting you sat in, the one that felt fine and produced nothing. I can already tell you which block was missing.

It's this one. Every time.

And the interesting part isn't that people forget to commit. It's that there's a structural reason rather than a personal one. Commit sits at the end. You are always out of time at the end. So the block that converts the entire day into action is the block that gets squeezed by every overrun that happened before it.

Which means this isn't a discipline problem you solve by caring more. It's a design problem you solve by protecting the time before you need it, which is module four's whole job.

08 Close

On screen CLOSE · "End deliberately"

Close. Reflect, acknowledge, and end on purpose. Not when the room booking runs out. On purpose.

Skip it and the session dribbles away, people drift out in twos and threes, and everyone leaves with a vague feeling that it didn't finish. Which it didn't.

And the cost isn't emotional, it's practical. The close is where people register what they just committed to. There's a real difference between having agreed to something in the flow of a busy afternoon and having heard it read back to you at the end. No close, no registration, and the commitments decay faster.

09 A real one, pulled apart

On screen a real seven-row agenda, unlabelled, then labelled

This is real, and I ran it. A strategy session for a product team, and I haven't tidied it up.

Take a second and label the rows in your head. Frame, diverge, converge, commit, close.

[pause]

Here's what I had. Welcome and context: that's a frame, and a thin one. Two ideation rounds, both diverge. A long discussion trying to be converge. A summary from me. And then, at the end, 'next steps to follow.'

Add it up. Fifty minutes of generating. Ten minutes of choosing. And a commit block that isn't a block at all, it's a promise to send an email.

Everybody left happy. I got good feedback on it. And nothing happened. Three weeks later somebody asked what we'd decided and there were four different answers.

That's the failure this module exists to stop. Not a bad workshop. A workshop that felt good and changed nothing, which is worse, because nobody learns anything from a session that felt fine.

10 The four joints, and the one that breaks most sessions

On screen the five blocks with the four gaps highlighted, DIVERGE ⇢ CONVERGE lit

Back to the chicken. The chef wasn't fast because he was strong. He was fast because he knew where the joints were.

Same here. Between every pair of blocks there's a handover, and the handovers are where sessions leak. Four joints: frame to diverge, diverge to converge, converge to commit, commit to close.

It's invisible while it's happening, which is what makes it dangerous. Nothing goes wrong loudly. The conversation just gets slightly worse and nobody can say why.

And one of the four breaks more sessions than the other three put together. The most common design fault in workshops, by a distance, is moving from diverge to converge without saying so out loud.

The group is still generating. You've already started deciding. And now, quietly, the two halves of the room are in different meetings.

Half of them are still thinking of new ideas and wondering when they'll get to put them up. The other half are trying to choose between what's there. Every time somebody adds one, the choosers get more frustrated, and every time somebody starts evaluating, the generators feel shut down.

It feels like the discussion's gone woolly. Actually, you just never told anyone the rules changed.

11 All four transitions, written out

On screen four sentences, one per joint

So every joint gets a line, and here are all four in the words I'd actually use. Steal them.

Frame to diverge: 'Right, that's the shape of the day. First job is to get everything on the table, so for the next twenty minutes we're collecting, not judging. Nothing's too obvious and nothing's too ambitious.'

Diverge to converge: 'We're done adding. Nothing new goes on the wall from here. Now we choose.'

Converge to commit: 'So that's the decision. Now the only question that matters: who's doing what, and by when.'

Commit to close: 'That's the work done. Last ten minutes is about how it went, and then we're finished.'

Four sentences. Between them they prevent most of the confusion in most of the workshops you'll ever run.

12 Kaner's Diamond and the Groan Zone

On screen the diamond, three zones, the middle band lit red

Now the most useful single diagram in facilitation. It's Sam Kaner's, from the Facilitator's Guide to Participatory Decision-Making, 1996, and if you read one thing after this path, make it that.

Read it left to right. On the left, the problem, and a group with its familiar opinions. On the right, a decision.

First the divergent zone. Options widen, more of the problem becomes visible than anyone brought in. Energy's generally good here. People enjoy this part, because it feels productive and it is.

On the far side, the convergent zone. Shared understanding, options combining and reducing, a group that can now actually decide something.

And then there's the middle. In the middle, the range is too wide to hold in one head. People are hearing views they don't share and can't yet see how any of it resolves. It feels like confusion, and repetition, and low-grade frustration, and everybody in the room wants out of it.

That is not the session failing. That's the session working.

Every group has to struggle to build shared understanding. There's no route around it, only through. And the struggle feels bad from the inside, always, including to you, including when you designed it deliberately and were sitting there expecting it.

13 What rescuing costs, and what to do instead

On screen the diamond with a dashed arc cutting over the middle, then three moves

What untrained facilitators do at this point is rescue the room.

Cut the discussion short. Offer their own synthesis, which is usually pretty good, because you've been listening for an hour and you genuinely can see it before they can. Or jump straight to a vote.

And the room is grateful. That's the trap. It feels like good facilitation, in the moment, to everyone present, including you.

But the decision doesn't hold. Six weeks later it quietly unwinds and nobody can point at why, because the thing that unwound was never really theirs. It was your decision wearing the group's clothes.

So what do you do instead. Three things, and they're all sentences.

Name it: 'This is the messy part. This is normal.'

Normalise it: 'Every group hits this. It's supposed to feel like this.'

And hold: 'Let's give it fifteen more minutes before we try to resolve anything.'

Then you stand there and let it be uncomfortable, which is the single hardest physical skill in this job. One trick that makes it survivable: show the group this diagram during your frame block, at the start of the day. A room that's been warned the middle will feel bad tolerates it several times longer than one that hasn't.

And the honest counter-argument, because I don't want you looking for a Groan Zone in every meeting you run. Not every session has one, and manufacturing struggle where there isn't any is its own failure. A retro on a sprint that went fine, a kickoff where everyone agrees on the goal: those have a diamond so shallow it isn't worth drawing.

Fair. The diamond scales with genuine disagreement. If the group actually agrees, converging is fast and that's a good day, not a failed one.

What I'd hold onto is this: check the agreement is real before you accept it. Cheap agreement in the first twenty minutes is very often six people deferring to the most senior view, and that looks identical to consensus from where you're standing. If you're not sure, run one silent written round and find out.

14 1-2-4-All

On screen 1 alone · 2 in pairs · 4 in fours · then all

One mechanic before we build. One minute alone, in writing, nobody speaks. Two minutes in pairs. Four minutes in fours. Then, and only then, share with the whole room.

It's called 1-2-4-All, it comes from Liberating Structures, and it's the closest thing this craft has to a universal tool.

It solves three problems at once. Everyone thinks before anyone speaks, so the quiet ones get their idea onto paper before the loud ones fill the air. Every idea gets pressure-tested in a pair, where it's cheap to be wrong, before it goes public. And a room of thirty contributes about as fully as a room of six would have.

If you take one mechanic from this entire path, take that one. It drops into almost any block of almost any session.

15 The kit, mapped to the blocks

On screen the five blocks, each with its exercises, hopes and fears filed under FRAME

Here's your handout. Ten exercises, and between them they cover all five blocks.

In frame: hopes and fears, and working agreements. In diverge: silent brainwriting, Crazy 8s, and how-might-we. In converge: affinity mapping, dot voting, and impact and effort. In commit: the commitment round. And in close: rose, thorn, bud.

That's the whole kit. You can build a complete, credible workshop knowing only these ten, and most professional facilitators run some version of most of them most weeks. The kit isn't the ceiling. It's the floor, and the floor is enough to start.

One that people misfile, and it's always the same one. Hopes and fears uses sticky notes and generates a lot of material, so it looks like divergence. It isn't. It's frame.

Its job is surfacing what the room walked in carrying, so you know what you're dealing with. It doesn't widen your option set on the actual problem at all.

Why that matters practically: if you file it as diverge, you'll count it towards your generating time, run twenty minutes of it, and then discover at half past eleven that you haven't actually started.

16 The practice

On screen the three agendas, then the Workshop Skeleton handout

Here's your one thing, and it's two reps. Give it a proper twenty minutes.

First rep, diagnosis. There are three real agendas on the handout, and I haven't tidied those up either. For each one: label the five blocks, name the block that's missing, and write the sentence you'd say at every joint. I'll tell you now that one of the three has no commit block at all, and one of them runs diverge and converge at the same time. If you find those two, you're reading it right.

Second rep, the skeleton. Take the outcome supplied on the handout. Check it's an artifact and not a feeling. Then build backwards from it. Ninety minutes, five blocks, timed, with real numbers in the boxes. At least one exercise from the kit in diverge, one in converge, one in commit.

And write all four transitions out in the words you'd actually say. Not the word 'transition'. Not 'move to voting'. The actual sentence, on the page, while you're designing. Because the moment you need a transition line is the moment the room is loudest and you're forty minutes behind, and what comes out instead is 'okay, so, um, maybe we should start looking at which of these', which is not a transition, it's a suggestion, and the room will treat it as one. Four minutes at the desk, and it's the highest return four minutes in your prep.

Two things I'd stop you on if I were standing next to you. If you're writing activities before you've finished the outcome box, go back, because the activities are the fun part and also the part that doesn't matter yet.

And when you're done, look at the number in your commit box. If it's under ten minutes, ask yourself who owns the first action and when it's due. If you can't answer from what you've written, your commit block is decoration.

Hang on to that sheet. It's the reference pattern you reach for every time you sit down to design a real agenda.

17 Callback close

On screen the chef's hands, a chicken in pieces

The chef wasn't fast because he was strong. He was fast because he knew where the joints were. He wasn't cutting through anything. He was finding the places where the thing already came apart.

Frame, diverge, converge, commit, close. And the four joints in between, where every session you've ever sat in was quietly leaking.

You know where they are now.

0:24 · then: Next: Module 4 - Designing the Agenda

A skeleton is not an agenda.

Next we put real time on it, and find out that everything takes about three times longer than you just wrote down.

## The practice (recap)

Two reps on Handout B. First, diagnose the three supplied agendas: label the five blocks, name the one that's missing, and write the sentence you'd say at every joint. One has no commit block; one runs diverge and converge at the same time. Then build a ninety-minute skeleton for the outcome supplied on the handout, checking first that it's an artifact and not a feeling. Five timed blocks, one kit exercise in each of diverge, converge and commit, and all four transitions written in the words you'd actually say.

## Callback

The chef wasn't cutting through anything. He was finding where the bird already came apart. Five blocks and four joints, in every session you'll ever run.

## Sources

Kaner, S. (1996). *Facilitator's Guide to Participatory Decision-Making*. New Society Publishers. Source of the diamond and the Groan Zone in Beats 14 and 15.

Lipmanowicz, H., & McCandless, K. *Liberating Structures*. Source of 1-2-4-All in Beat 16.

The block proportions in Beat 3 are mine, from practice. They aren't from a study and shouldn't be presented as one.