06 · Difficult Conversations¶
The conversations you postpone are the ones that define you as a leader, because the whole team can see what you tolerate. The delay is never kindness — the person carries on doing the thing, the gap between their self-image and your view widens for months, and when it finally surfaces it arrives as a shock and often as an unfair one. "Why is this the first I'm hearing of it?" is a legitimate complaint about a real management failure. This module gives you a preparation worksheet, an opening script structure that avoids the two standard traps, and two worked conversations — one on behaviour, one on performance.
1. Why the opening sentence decides the outcome¶
Almost all difficult conversations fail in the first thirty seconds, in one of two directions.
| Trap | Sounds like | What the other person hears |
|---|---|---|
| The soft open | "How's everything going? Great. Anyway, small thing…" | Nothing is wrong. They leave unaware there was a message |
| The verdict open | "We need to talk about your attitude in meetings." | I am on trial and the judgement is already made. Defence mode |
| The sandwich | Praise, criticism, praise | The praise is a delivery mechanism, so future praise is now worthless |
The alternative is what Stone, Patton and Heen call the third story: open with a description a neutral observer would recognise, which contains your concern but not your conclusion, and then invite their version before you give yours.
- Verdict: "You've been dismissive of Rina in the last three meetings."
- Third story: "I've noticed something in the last three planning meetings that I'd like to compare notes on — a few times Rina started a point and it didn't land, and I want to understand what's happening there from where you sit."
The second version says the same thing and leaves the conversation open. It is not softer; it is more precise.
2. The preparation worksheet¶
Do not improvise these. Twenty minutes of preparation is the difference between a conversation and an argument.
DIFFICULT CONVERSATION PREP
1. FACTS (only what a camera would have recorded)
Specific instances, with dates:
What I actually observed vs. what I inferred:
2. IMPACT
The concrete cost — to the work, to other people, to them:
How do I know that impact is real and not my assumption?
3. THEIR STORY (write it as if you were them, generously)
What might make their behaviour completely reasonable?
What pressure might they be under that I can't see?
What might they say that would change my mind?
4. MY CONTRIBUTION (there is nearly always one)
What did I fail to say, and when should I have said it?
What have I rewarded that I'm now criticising?
5. OUTCOME
What must be different, specifically and observably:
What is negotiable, and what genuinely is not:
6. OPENING SENTENCE (write it out, word for word)
______________________________________________
7. LANDMINES
What might they say that would knock me off course?
My response if they cry / get angry / go silent:
Box 3 is the one people skip and the one that changes conversations. If you cannot construct a version of events in which the person is behaving reasonably, you have not prepared — you have rehearsed.
3. The conversation structure¶
Six moves, in order. The order is the method.
| Move | Purpose | Wording |
|---|---|---|
| 1. Signal | Remove ambiguity about the register | "I want to raise something that's been on my mind. It's not a crisis, and I'd rather say it than sit on it." |
| 2. Third story | Put the observable on the table | "In the last three planning meetings, I noticed…" |
| 3. Ask first | Get their version before you conclude | "What's your read on it?" — then actually listen (Level 1, Module 3) |
| 4. State impact | The cost, not the character | "The effect is that we lose the objection early and pay for it late." |
| 5. Own your part | Restore symmetry | "I should have said this after the first meeting, not the third." |
| 6. Agree the change | Specific, observable, dated | "What would you do differently on Tuesday?" |
The one non-negotiable: move 3 comes before move 4. If you deliver impact before hearing their version, everything after it is a defence of your judgement rather than a conversation about the work.
4. A worked conversation: behaviour¶
Rina is quiet in planning meetings. Tom, the most experienced person on the team, talks over her — Tom is well-regarded, works hard, and has no idea he is doing it. Their manager, Yusuf, has prepared the worksheet.
Yusuf: I want to raise something that's been on my mind for a couple of weeks. It's not a crisis, and I'd rather say it than sit on it. In the last three planning meetings I've watched Rina start a point and not finish it — twice you came in over the top of her, and once you summarised what she was saying before she got there. I want to compare notes, because I don't know what it looks like from where you're sitting.
Tom: Really? I— honestly, I thought I was helping. She takes a while to get to the point and we're always short on time.
Yusuf: Say more about the time pressure.
Tom: We've got 45 minutes and eleven items. I'm trying to keep it moving. And when Rina's uncertain about something she prefaces it a lot, so I jump in to speed it up.
Yusuf: That's useful — I'd assumed it was impatience with her and it sounds like it's impatience with the agenda. Here's the effect I'm seeing, though. In the last two meetings, Rina raised a concern about the regional rollout dates that got cut off. She was right. We found out three weeks later, from the region, and it cost us a fortnight. The thing she was slow to say was the thing we needed most.
Tom: ...I didn't know that's what she'd been saying.
Yusuf: And I should have raised this after the first meeting rather than the third. Some of this is my agenda — eleven items in 45 minutes is my design, and it's creating the pressure you're responding to. I'll cut that to six.
Tom: That would genuinely help.
Yusuf: What would you do differently on Tuesday?
Tom: Not interrupt, obviously. But if I just sit on it I'll be counting the seconds. Could I ask her directly at the point where she matters — "Rina, what's the regional view?" — rather than waiting and then jumping in?
Yusuf: That's better than what I was going to suggest. Do that. I'll watch on Tuesday and tell you afterwards what I saw, either way.
Yusuf changed his own meeting design in move 5, which is why Tom could accept move 4 without defending himself. He also got a better solution than the one he'd prepared, because he ran move 3 properly.
5. A worked conversation: performance¶
Different case — repeated missed commitments, third occurrence, and the message must be unmistakable. Here the third story does not mean softening the standard.
Yusuf: This is a serious conversation and I want you to hear it clearly. The monthly regional pack has been late three months running — the 9th, the 11th and the 12th, against a due date of the 5th. Each time I heard about it after the deadline, from someone else. What's going on?
Marc: The inputs from the two new regions come in late. I've been chasing them.
Yusuf: How long has that been true?
Marc: Since they onboarded. About four months.
Yusuf: Then I want to separate two things, because they need different fixes. The input delay is a real problem and I'll help with it — I can go to the regional leads this week. The part that's yours is that for four months you knew there was a structural problem, and you didn't tell me. Three times I found out after the fact. That's the thing that has to change, and it isn't negotiable.
Marc: I thought I could fix it before it became a thing.
Yusuf: I understand the instinct, and I'd rather have the flag. So, specifically: the pack is due on the 5th. If it's at risk, I hear about it on the 1st, from you, with what you need. If the inputs are still the blocker after I've talked to the regional leads, we change the due date rather than miss it. If the next two months are late and unflagged, we're in a different conversation about the role, and I want to be straight with you that that's where this goes.
Marc: Understood. The 1st.
Yusuf: I'll put those dates in writing today so neither of us is working from memory. And I'll come back to you on Friday about the regional leads — that's my commitment in this.
Two things to copy. Yusuf separated the systemic cause from the behavioural one, so Marc could not use the real input problem to absorb the whole conversation. And he named the consequence explicitly — a difficult conversation in which the person leaves unaware that their job is at stake has not been had.
6. Landmines¶
- Tears or anger. Pause, don't retract. "Take a minute. I'm not going anywhere, and I'm not withdrawing what I said."
- "Who told you this?" Speak only to what you observed yourself, and if you can't, don't raise it until you can.
- Counter-attack. Note it, park it, finish yours. "That's a fair thing to raise and I want to hear it properly — let's finish this first, then I'll take it seriously."
- Total agreement. Suspiciously fast agreement usually means escape. Ask "what will you actually do differently?" and listen for specifics.
How It Actually Works¶
Why the opening sentence decides the outcome is explained by a concept from conversation-analysis research called preference structure: the first turn of an exchange sets up which of several possible next moves the other party can produce without it counting as a socially costly deviation. An opening like "we need to talk about your performance" structurally invites either silent acceptance or open confrontation as the two "fitting" next moves — there's no comfortable slot for the collaborative, curious response a leader might actually want. An opening that names the specific behavior and observed impact, then asks a genuine question ("here's what I noticed and its effect — help me understand what was going on"), creates a structural slot for exactly that collaborative response, because the turn itself was built to invite it. This is the same SBI logic from the Level 1 feedback module, applied to a higher-stakes conversation.
Why preparation worksheets outperform winging it, specifically for difficult conversations. Under the anticipatory stress of a hard conversation, working memory capacity is measurably reduced (a well-studied effect called stress-induced cognitive narrowing), which means a leader who hasn't pre-loaded their key points, the behavior-impact facts, and their own emotional triggers into the conversation plan is trying to generate all of that live, under exactly the conditions least suited to generating it well. Writing it beforehand offloads that cognitive burden to before the stress spike, leaving more working-memory capacity during the actual conversation for listening and adapting — the same offloading principle behind pilots' pre-flight checklists.
Why landmines (the specific phrases and moves that derail these conversations) are worth memorizing explicitly rather than trusting good intent in the moment. Phrases like "you always" or "everyone thinks" trigger the identity-threat response covered in the feedback and psychological-safety modules almost automatically, regardless of the speaker's actual intent — the listener's threat-detection system responds to the absolute/generalizing language pattern itself, not to the specific content. Because this response fires before deliberate processing catches up (the same 100-200ms head start described in the self-awareness module), a leader's genuinely good intent doesn't prevent the phrase from landing as an attack; only avoiding the trigger pattern in the first place does.
Exercise¶
Name the conversation you have been avoiding — the one that surfaced in your mind while reading this. Complete the worksheet in section 2 for it, in full, including box 3 and box 4. Write your opening sentence word for word and read it aloud; if it contains a verdict or a diagnosis of character, rewrite it as an observation until it doesn't.
Then schedule it, inside seven days. Afterwards, write down two things: what you learned in move 3 that you did not know, and whether you got to move 6 with a specific dated commitment or settled for general agreement.