09 · Servant Leadership in Meetings¶
Meetings are where your leadership is most visible and least examined. Everything from Levels 1 and 2 — listening, safety, conflict, coaching — either shows up in the room or is theoretical. The default pattern is well known: the most senior person states a view first, the same two people speak, everyone else calibrates to what has already been said, and the meeting produces agreement that dissolves the moment people leave. Serving the group's thinking means designing the meeting so the information in the quietest heads reaches the decision. This module gives you a meeting design canvas, a facilitation move set, and a worked before-and-after of the same meeting run two ways.
1. One meeting, one job¶
Most bad meetings are two or three meetings wearing a trenchcoat. Name the type before you write the agenda, and never run two types in one block without a hard break between them.
| Type | Output | Right format | Fatal mistake |
|---|---|---|---|
| Inform | Shared understanding | Written first, meeting only for questions | Reading slides aloud to people who can read |
| Generate | A wide set of options | Silent generation, then build | Starting with open discussion — first idea anchors everything |
| Decide | A named decision with an owner and date | Options pre-circulated, decision rights stated up front | Discovering mid-meeting that nobody knows who decides |
| Build | Alignment and commitment | Small group, real work done in the room | Inviting an audience |
| Connect | Relationship and trust | Unstructured, protected time | Adding an agenda and killing the point |
The single highest-value question to ask before any meeting you call: what will be true at the end that isn't true now? If you can't answer it in one sentence, cancel and send a note.
2. The meeting design canvas¶
MEETING DESIGN CANVAS
Type (one only): inform / generate / decide / build / connect
What is true at the end that isn't true now:
____________________________________________
DECISION RIGHTS (state these out loud at the start)
Who decides:
Who must be consulted (their input changes the outcome):
Who is informed (present for context, not for a vote):
Is this actually already decided? yes / no
If yes, this is an INFORM meeting. Say so and don't fake a debate.
ATTENDEES
Name — why they are here — what happens if they aren't:
(No answer to column three = remove them and send the notes)
PRE-WORK
What is circulated, and when (24h minimum):
What people should arrive having done:
DESIGN
First 5 minutes: (silent read / silent write / round)
How the quiet people get airtime:
When I speak: (target: last, on any question I own)
How dissent is invited: (specific mechanism, not "any concerns?")
CLOSE
Decision, owner, date — read aloud before anyone leaves:
What I will send in writing, and by when:
Who was not in the room and needs to hear it from me directly:
3. Facilitation moves¶
Techniques with mechanisms, not exhortations.
| Move | How | What it fixes |
|---|---|---|
| Speak last | Withhold your view until everyone with a view has given theirs | Anchoring on the senior voice — the biggest single distortion in the room |
| Silent start | Five minutes reading the doc or writing options, in the room, no talking | Meetings where only the fast talkers and the pre-prepared contribute |
| Brainwriting | Everyone writes three options privately before any discussion | First-idea anchoring in generate meetings |
| Round | Go round the table by name, everyone says one thing, no cross-talk | The same two people occupying 80% of airtime |
| Named dissent | "I want the strongest argument against this — Ravi, you go first" | Groupthink; makes objecting the assigned task rather than a risky act |
| Airtime count | Tally who spoke, privately, for one meeting | Your own inaccurate sense of how balanced the discussion was |
| Process, not content | "We've been on this ten minutes with no new information — decide or park?" | Meetings that drift and end at the time limit rather than at an answer |
| Read-back close | State decision, owner and date aloud before anyone leaves | Meetings that "agreed" and produce nothing |
Speak last is the whole module in one habit. If you state your view first, the rest of the meeting is a referendum on you, and you will mistake compliance for consensus.
4. Same meeting, two ways¶
A decision meeting: whether to move customer onboarding from the sales team to the support team. Convener is Dara, the director. Present: Ines (sales lead), Kwame (support lead), Bea (ops analyst), Tunde (new regional manager, four weeks in).
Version one, the default
Dara: Thanks all. So — I've looked at this and I think onboarding clearly belongs in support. Sales shouldn't be doing implementation. Thoughts?
Kwame: Agreed, makes sense.
Ines: I mean — if that's the direction, we'd need headcount to move with it, but yes, I can see it.
Dara: Bea?
Bea: No objections from me.
Dara: Tunde?
Tunde: Happy to go with the room.
Dara: Great, that was easy. Let's do it.
Eight minutes, unanimous, and worthless. Dara anchored in sentence one, so nobody after her was giving a view — they were responding to hers. Bea has churn data showing onboarding quality drives first-year renewal. Tunde has watched this exact move fail at his previous employer. Neither piece of information reached the decision.
Version two, designed
Dara: Decision today: where onboarding sits from Q3. To be clear up front about rights — I decide, but Ines and Kwame's input changes the outcome because it's their headcount, and Bea and Tunde are here because they know things I don't. This is not pre-decided; if it were I'd tell you.
Dara: Five minutes, silently, on the doc I sent. Write down the single strongest argument for each option, including the one you don't favour. Then we'll go round.
(Five minutes.)
Dara: Round. Tunde first — you're newest, so you're least contaminated by how we've always done it.
Tunde: At my last company we made this exact move. It failed, and the reason was specific: support was measured on ticket closure time, and onboarding is slow work. Within two quarters they'd deprioritised it into nothing. The structure wasn't the problem, the metric was.
Bea: Mine's the other direction. Accounts with an onboarding quality score above 7 renew at 91%; below 4, it's 62%. That's the largest single lever in the renewal number, and right now nobody owns that score.
Kwame: My argument for taking it: my team already picks up the failures at month three, blind, with no context. My argument against: Tunde is right about the metric, and I'd be signing up for work my current targets punish me for doing.
Ines: My argument for keeping it: relationship continuity through the first 90 days. My argument against — and this is uncomfortable — my reps are compensated on new bookings, so onboarding is the thing they do last. I'd be defending a structure my own comp plan undermines.
Dara: Now mine, and you'll notice it's changed. I came in convinced this was a structural question. Three of you have told me it's a measurement question — both teams are incentivised against doing it well, so moving it just relocates the problem.
Dara: Strongest argument against moving it, then. Kwame, you go — you're the one who'd gain the team.
Kwame: That we spend two quarters on a reorg and land exactly where we are, with a worse handover.
Dara: That's persuasive. Here's the decision: onboarding moves to support from Q3, and it moves with a dedicated onboarding quality metric that is not ticket closure time — Bea's score, owned by Kwame, reported alongside renewal. Ines, two of your onboarding specialists move with it; I'll fund the third from my budget rather than take it from your quota headcount. Owner of the transition plan: Kwame, with Bea. First draft by the 20th.
Dara: Read-back so nobody leaves with a different version. Move to support, Q3, new metric replaces closure time for that work, two heads move plus one funded by me, Kwame owns the plan, draft on the 20th. Anything I've got wrong? ... I'll send that in writing today, and I'll talk to the two specialists myself before it's announced.
Twenty-five minutes instead of eight, a materially better decision, and the person four weeks into the company supplied the insight that changed it — because he was asked first, before anyone senior had spoken.
5. Meetings you attend but don't run¶
Servant leadership in someone else's meeting is mostly restraint plus one or two well-placed moves.
- Give your view late even when you're not chairing. Your seniority anchors regardless of who called the meeting.
- Redirect credit and questions. "That analysis is Bea's, and she's in the room" costs you nothing and changes what happens next.
- Name the process, not the content. "Can I check who's deciding this?" is the single most useful sentence available to a participant.
- Amplify a cut-off point. "I want to come back to what Rina started to say" — a five-word intervention that changes who gets heard.
How It Actually Works¶
"One meeting, one job" works because it removes a well-documented cause of group dysfunction called process loss — the gap between a group's theoretical combined capability and what it actually produces, caused by coordination overhead rather than any individual's fault. A meeting that mixes information-sharing, decision-making, and open brainstorming forces every participant to constantly re-detect which mode is currently active (is this the moment to share a concern, or to just listen?), and that mode-detection overhead consumes attention that would otherwise go toward the actual content. A single-purpose meeting removes the mode-detection tax entirely — participants know from the invitation which cognitive mode to arrive in, which is why single-purpose meetings routinely run shorter and produce clearer outcomes even holding total content constant.
Why facilitation moves that redistribute airtime aren't just fairness — they change what information the group has access to. Group discussion research consistently finds that verbal participation correlates weakly at best with the quality or relevance of a person's information, and that the loudest or most senior voices systematically dominate airtime regardless of who actually holds the most relevant information for the decision at hand (a pattern sometimes called "information pooling bias" — groups over-discuss information everyone already shares and under-surface information only one person holds, precisely because the person who holds it uniquely often doesn't get the airtime to say it). Deliberate facilitation moves — round-robin input, explicitly inviting the quietest person before the most senior one weighs in — are a direct structural countermeasure to this bias, not a politeness gesture.
Why "meetings you attend but don't run" require the same discipline mechanism as delegation. As a participant rather than the facilitator, the temptation to answer a directed-to-someone-else question first, or to resolve ambiguity before the actual owner gets a chance to, replicates the low-delegation dynamic from Level 1 at meeting scale: it signals (correctly, over enough repetitions) that the fastest path to an answer in this room is to wait for the senior person, not to think it through themselves, training exactly the passivity a servant leader is trying to avoid.
Exercise¶
Take the next meeting you own and fill in the canvas in section 2 before you send the invite. Force yourself through two boxes honestly: for each attendee, what happens if they aren't there; and whether the thing is already decided. If it is, convert the meeting to an inform and say so out loud — faking a consultation is one of the fastest ways to lose the trust you built in Level 1.
In the meeting itself, use exactly two moves from section 3: speak last on any question you own, and one of silent start, round, or named dissent. Keep a private airtime tally as you go. Afterwards, write down who spoke least, what they said when they finally did, and whether the outcome differed from what you would have decided in the first two minutes. If it never differs, you are chairing a ratification process rather than a decision.