Script Any Difficult Workplace Conversation Before You Walk In
Why this prompt matters
A poorly delivered difficult conversation rarely just fails once — it tends to recur, and each repeat costs more. Gallup has tied disengaged, poorly-managed teams to roughly 18% lower productivity and significantly higher voluntary turnover; replacing a mid-level technical employee typically runs 50-200% of their annual salary in recruiting, onboarding, and lost-productivity costs. Beyond the direct cost, an avoided or mishandled confrontation lets the underlying behavior calcify — the blame-shifting in this scenario, left unaddressed, tends to spread to other team members within a quarter, and by the time it reaches HR it's a formal complaint rather than a coaching conversation. Managers who wing these conversations also disproportionately trigger the two worst outcomes: the employee feels blindsided (which reads as unfair and drives grievance, even litigation risk in termination scenarios) or the manager backs off mid-conversation and nothing changes at all.
What we use it for
You're a mid-level engineering manager and you have a 1:1 scheduled for tomorrow morning with a senior team member who has missed three consecutive sprint deadlines and has started publicly blaming other teams for the delays in front of the group. You've never had to confront someone senior about both a performance pattern and a trust-damaging behavior in the same conversation, and you're worried about either being too soft (letting the pattern continue) or too harsh (damaging a working relationship you still need).
Prompt
Act as an experienced workplace mediator and executive communication coach who has prepared hundreds of managers for high-stakes conversations, including layoffs, performance confrontations, and conflict resolution. Context: I need to have a difficult conversation at work. Here are the details: - Who I'm talking to: [THE PERSON YOU'RE TALKING TO — name/role] - My relationship to them: [YOUR RELATIONSHIP TO THEM — e.g., direct manager, peer, skip-level] - The situation: [THE SITUATION — describe what happened and what needs to be addressed] - What makes this sensitive: [ANYTHING THAT MAKES THIS SENSITIVE — e.g., prior history, emotional stakes, legal/HR considerations, power dynamics] - What outcome I want: [WHAT OUTCOME YOU WANT — the ideal result of this conversation] - When this is happening: [TIMING — e.g., tomorrow morning, end of this week] Task: Write a complete conversation script I can use to prepare for and guide this conversation in real time. Constraints: - Keep the tone honest and direct, never falsely soft or euphemistic — avoid corporate jargon that obscures the message. - The opening line must state the purpose within the first two sentences; do not bury the point. - Anticipate realistic pushback based on human psychology (defensiveness, denial, anger, bargaining), not generic objections. - Every prepared response must acknowledge the pushback before redirecting — no dismissiveness. - Keep the entire script deliverable in under 10 minutes of actual talk time. - Do not suggest recording the conversation or bringing in a witness unless the situation explicitly involves formal HR/legal process. Output Format: 1. **Opening Line** (1-3 sentences, exact wording to say first) 2. **Key Points** (3-5 bullet points, the core message in the order to deliver it) 3. **Anticipated Pushback & Responses** (2-4 pairs: the likely reaction, and exactly what to say back) 4. **Closing & Next Steps** (how to end the conversation and what happens after) 5. **One-Line Reminder** (a single sentence to keep in mind if the conversation goes off script)
Result
Scenario used for this example: You are a mid-level engineering manager. A senior developer on your team, Priya, has missed three consecutive sprint deadlines and has started publicly blaming other teams in stand-ups for the delays, which is damaging cross-team trust. You need to address the pattern directly, tomorrow morning, in a 1:1.
1. Opening Line
"Priya, I want to talk about the last three sprints and how we're communicating about delays with other teams — this isn't a performance review, it's a conversation I need to have now before it becomes a bigger problem."
2. Key Points
- The team has missed three consecutive sprint commitments, and the pattern is now visible to stakeholders outside the team.
- In the last two stand-ups, the delays were attributed to the platform team and the design team without first checking with them — that's created friction I've had to smooth over twice this week.
- I need to understand what's actually blocking delivery — whether it's scope, dependencies, or something else — so we can fix the root cause together.
- Going forward, any blocker involving another team gets raised to me first, so we can verify it and raise it jointly instead of attributing blame in a group setting.
- This is fixable, and I'm not looking to punish you for it — I need us to reset how we handle blockers as a team.
3. Anticipated Pushback & Responses
- Pushback: "The platform team really has been slow, I'm not wrong about that."
Response: "You might be right that they're a bottleneck — I'm not disputing the facts, I'm addressing how it was raised. Calling it out in stand-up in front of five other people isn't the way to solve it. Bring it to me and we'll escalate it properly." - Pushback: "It feels like I'm being singled out — other people miss deadlines too."
Response: "This isn't about singling you out, it's about a three-sprint pattern with your name on it and its effect on other teams. If I see the same pattern elsewhere, I'll have this same conversation there." - Pushback: (goes quiet, disengages)
Response: "I know this isn't a fun conversation to be in. Take a moment — I want to actually hear what's been getting in the way, because I'd rather solve it than keep having this talk."
4. Closing & Next Steps
"Let's agree on two things before you go: first, any inter-team blocker comes to me before it comes up in stand-up. Second, let's set a check-in this Friday to see where the current sprint stands. I'll send a two-line summary of what we agreed on right after this so we're both working from the same page."
5. One-Line Reminder
"Stay on the pattern and the impact, not on Priya's character — if the conversation drifts into 'you always' or 'you never,' pause and come back to specific sprints and specific stand-ups."
Most advice about difficult conversations tells you what to feel — stay calm, be empathetic, listen actively. It rarely tells you what to actually say when the other person pushes back, goes quiet, or gets defensive in the room. This prompt closes that gap: instead of generic advice, it produces a word-for-word script built around your specific situation.
Why this prompt is structured the way it is
The prompt opens by casting the AI as a mediator and executive coach rather than a generic assistant. That framing matters — it pulls the model toward direct, field-tested phrasing instead of the softened, hedge-everything language a general-purpose assistant defaults to when a topic sounds sensitive.
The context block asks for five specific inputs: who you're talking to, your relationship to them, the situation, what makes it sensitive, and the outcome you want. Each of these changes the script materially. A conversation with a peer needs different framing than one with a direct report or a skip-level, and "what makes this sensitive" is what lets the model account for history — prior conflict, protected-class considerations, or emotional volatility — that a generic template would miss entirely.
The constraints section is doing the real quality-control work. Requiring the opening line to state the purpose within two sentences prevents the single most common failure mode in hard conversations: burying the point under five minutes of small talk, which makes the eventual pivot feel like an ambush. Requiring every prepared response to acknowledge pushback before redirecting stops the script from producing responses that sound dismissive or scripted when the other person reacts emotionally — which, in a real difficult conversation, they usually do.
The five-part output format mirrors how these conversations actually unfold in real time: an opening you say verbatim, key points in the order that keeps the conversation from derailing, prepared responses to the two or three reactions you can actually predict, a closing that locks in next steps instead of trailing off, and a one-line reminder to pull yourself back on track if things go sideways — because scripts rarely survive contact with an emotional room intact.
Where this earns its keep
This isn't limited to layoffs or terminations. It works for the more common and equally uncomfortable cases: telling a peer their behavior in meetings is undermining you, giving a long-tenured report blunt feedback that their skills haven't kept pace with the role, addressing a conflict between two team members who both report to you, or pushing back on a client who keeps changing scope without adjusting the budget. The pattern — state it early, expect specific pushback, respond without being dismissive, close with concrete next steps — holds across all of them.
Run it the night before, not five minutes before. The value isn't just the script itself; it's that writing out the anticipated pushback forces you to think through the conversation's hardest moment before you're standing in it.