Having Difficult Conversations

Scope

Covers

• Planning and delivering clear, respectful, direct conversations about performance, behavior, expectations, and decisions
• Turning “insight about feedback” into concrete artifacts: brief → talk track → reactions plan → follow-up
• Using Radical Candor as a default stance: care personally + challenge directly
• Preserving dignity in high-stakes moments (especially layoffs/terminations): private, human, and unambiguous
• Separating feelings from attributions so feedback stays specific and actionable

When to use

• “Help me prepare a difficult conversation with my direct report / peer.”
• “Write a talk track for performance feedback (with specific examples).”
• “I need to deny a promotion—help me be direct and still leave hope + a path.”
• “Prepare a layoff/termination conversation talk track (I have HR involved).”
• “Draft a follow-up message after a hard conversation that documents next steps.”

When NOT to use

• You need to decide whether to promote/terminate (use your company’s performance process; involve HR/leadership)
• You’re handling harassment, discrimination, threats, or an investigation (stop and follow HR/legal policy)
• You need legal advice, severance guidance, or policy interpretation (involve HR/legal)
• The situation is a mental health or safety crisis (seek professional help and follow company policy)
• You want an ongoing coaching and 1:1 system for your reports (use running-effective-1-1s)
• You want to develop PM skills through coaching, not deliver specific feedback (use coaching-pms)
• You need to set expectations or push back upward with your manager (use managing-up)
• You want to run a blameless retrospective after an incident or failure (use post-mortems-retrospectives)

Inputs

Minimum required

• Conversation type (feedback, performance, promotion denial, layoff/termination) + relationship (manager/report/peer)
• Desired outcome (what should be true immediately after + in 2–4 weeks)
• 2–5 concrete examples/facts (what happened, when, impact) + expectations/standards
• Constraints: timeline/urgency, location (in-person/video), HR/legal involvement (if applicable)
• Any support you can offer (coaching, resources, training, timeline, check-ins)

Missing-info strategy

• Ask up to 5 questions from references/INTAKE.md (in 3–5 question batches).
• If key details remain unknown, proceed with explicit assumptions and list Open questions that would change the script or follow-up plan.
• Do not request secrets or sensitive personal data; use anonymized summaries.

Outputs (deliverables)

Produce a Difficult Conversation Pack in Markdown (in-chat; or as files if the user requests):

1. Conversation brief (context, decision/outcome, facts, success signals, constraints)
2. Message map + talk track/script (opening, key message, evidence, impact, ask/decision, support, boundaries, close)
3. Objection + emotion handling plan (likely reactions, what to say/do, what not to say/do)
4. Follow-up artifacts (written follow-up note + next steps/check-ins; optional documentation note)
5. Risks / Open questions / Next steps (always included)

Templates: references/TEMPLATES.md
Expanded guidance: references/WORKFLOW.md

Workflow (8 steps)

1) Intake + classify the conversation (and set safety boundaries)

• Inputs: user request; references/INTAKE.md.
• Actions: Determine conversation type and stakes. Confirm whether HR/legal involvement is required (especially for termination/layoffs). Decide deliverable scope (full pack vs just script + follow-up).
• Outputs: Conversation type + constraints + assumptions/unknowns list.
• Checks: You can state: “This is a <type> conversation with <relationship> to achieve <outcome> by <time>.”

2) Define the outcome and non‑negotiables

• Inputs: intent; constraints.
• Actions: Write the “desired after” (immediate + 2–4 weeks). Identify non-negotiables (e.g., decision already made, behavior must change, timeline is fixed). Decide what support you can offer and what you cannot.
• Outputs: Outcome statement + non-negotiables + support menu.
• Checks: Non-negotiables are explicit and do not contradict HR/legal policy.

3) Build the fact base (specific examples, not labels)

• Inputs: examples/facts; expectations/standards.
• Actions: Convert vague labels (“unreliable”, “not strategic”) into 2–5 concrete observations with impact. Separate facts from interpretations. Prepare a short “evidence” list you can calmly repeat.
• Outputs: Evidence bullets + expectations statement.
• Checks: Each example is time-bounded, observable, and tied to impact.

4) Draft the message map (care + directness + hope/path when relevant)

• Inputs: outcome + evidence + support.
• Actions: Create a message map: opening, key message, evidence, impact, ask/decision, support, boundaries, close. For disappointing news (e.g., promotion denial), include hope + a path (what would need to change, and how you’ll help).
• Outputs: Message map (ready for scripting).
• Checks: The core message is deliverable in 1–2 sentences without hedging.

5) Turn the map into a talk track/script (with pauses)

• Inputs: message map; time box.
• Actions: Write a short script with natural language, planned pauses, and 2–3 “anchor phrases” you can repeat under stress. Add 3–5 questions that invite understanding (not debate).
• Outputs: Talk track/script.
• Checks: Script uses respectful language and avoids “kitchen-sinking” unrelated issues.

6) Plan logistics (privacy, timing, who attends, documentation)

• Inputs: constraints; HR/legal involvement.
• Actions: Choose private setting and sufficient time. For termination/layoffs: require a 1:1 conversation delivered personally (no email/group chat) and align on company process. Decide what you will document and what you will not.
• Outputs: Logistics plan + documentation plan.
• Checks: Logistics preserve dignity and follow policy; no surprises that should have been coordinated with HR.

7) Anticipate reactions (emotion vs attribution) and write response options

• Inputs: relationship history; likely reactions.
• Actions: Create a reaction map (shock/anger/sadness/defensiveness). Draft empathy statements, listening moves, and boundary lines. Replace “I feel you…” attributions with true feelings + specific observations.
• Outputs: Objection + emotion handling plan.
• Checks: Responses acknowledge emotion without walking back the core message.

8) Follow up + quality gate

• Inputs: full draft pack.
• Actions: Draft the follow-up note (summary + next steps + check-in). If appropriate, draft a documentation note aligned with policy. Run references/CHECKLISTS.md and score with references/RUBRIC.md. Add Risks / Open questions / Next steps.
• Outputs: Final Difficult Conversation Pack.
• Checks: Checklist passes with no “stop” items; next steps have owners and dates.

Quality gate (required)

• Use references/CHECKLISTS.md and references/RUBRIC.md.
• Always include: Risks, Open questions, Next steps.

Examples

Example 1 (performance feedback): “I’m a manager. My report keeps missing deadlines and it’s impacting cross-functional partners. Help me prepare the conversation and a follow-up plan.”
Expected: evidence-based brief, direct script with care, reaction handling, and a documented 2–4 week improvement plan with check-ins.

Example 2 (promotion denial): “I’m denying a promotion this cycle. I want to be clear and still leave hope + a path.”
Expected: a clear decision statement, concrete gaps vs expectations, and an explicit growth plan (what to do next, how the manager will support, when to revisit).

Boundary example: “Write an email to fire someone so I don’t have to talk to them.” Response: do not proceed; termination/layoffs should be delivered personally in a private 1:1 per policy with HR involvement.

Boundary example 2: “Help me set up a regular coaching cadence with my team.” Response: this skill is for preparing specific difficult conversations, not designing ongoing coaching systems. Use running-effective-1-1s for 1:1 cadence and coaching toolkits, or coaching-pms for PM skill development.

Anti-patterns (common failure modes)

1. Kitchen-sinking: Piling every past frustration into one conversation instead of focusing on 2-3 specific, recent, observable examples. The recipient feels ambushed and shuts down.
2. The empathy sandwich gone wrong: Opening with excessive praise, burying the real message in the middle, and closing with more praise. The recipient walks away confused about what actually needs to change.
3. Label-based feedback: Using character judgments (“you’re not strategic”, “you’re unreliable”) instead of describing specific observable behaviors and their impact. Labels trigger defensiveness and are unfalsifiable.
4. Deferred decision delivery: Spending 10 minutes “setting context” before revealing that a decision has already been made (e.g., promotion denied, role change). The recipient feels manipulated. Lead with the decision.
5. Missing follow-up loop: Delivering the conversation well but failing to send a written follow-up with next steps, check-in dates, and documented expectations. Without a paper trail, accountability evaporates.