name: difficult-conversation-preparation-runner description: > Executes the full Difficult Conversation Preparation SOP — from writing the issue brief through building a conversation script with anticipated reactions, fallback positions, and a scope creep response if applicable. Run when a sensitive topic needs to be raised with a client. metadata: author: "Kathryn Brown, Practice Builders" version: "1.0.0" date: "2026-04-28" sop: "Difficult Conversation Preparation" category: "Client Communication" frequency: "Trigger-Based" estimated-time: "30 min" trigger: "When a sensitive topic needs to be raised with a client"
Difficult Conversation Preparation — Runner
You are executing the Difficult Conversation Preparation SOP for an independent consultant. Walking into a difficult client conversation without preparation is how minor friction becomes a damaged engagement. This runner gets you into the conversation with a clear position, a fallback, and language that holds the relationship while it holds the line.
Do not skip steps. Do not ask questions across multiple turns — collect everything upfront.
What you'll have when this is done: A written issue brief, prepared opening language, a clear desired outcome, at least one acceptable fallback position, and — if applicable — a scope response ready to introduce. You walk in prepared, not reactive.
Step 1: Collect All Inputs
Gather the following from the user in a single prompt. Accept whatever detail level they provide. Flag gaps but keep moving.
The issue:
- One-paragraph description of what happened, what the expected state was, and what needs to be resolved (factual — no blame, no tone)
- Is this a one-time issue or part of a pattern? (first occurrence / second time / recurring)
Client relationship context:
- Client name and primary contact name
- Relationship tenure (how long you've worked together)
- Current relationship temperature (strong, neutral, strained)
- Client communication style (formal, casual, mixed, conflict-avoidant)
- Any prior difficult conversations with this client? (yes / no) — if yes, brief summary of how they went
Engagement context:
- Relevant SOW, change order, or prior communication thread (or summary of what was agreed)
- Current milestone progress and payment standing
- Is scope at issue? (yes / no)
- If scope is at issue: the specific out-of-scope work being requested and the relevant SOW section
Desired outcome:
- What you want to walk out of the conversation with (be specific)
- What you'd accept as a fallback if the client pushes back
- Is there a walk-away line — a minimum below which the engagement can't continue?
Step 2: Write the Issue Brief
Using the inputs, write a one-paragraph issue brief that covers:
- What happened — the factual trigger, no interpretation
- What the expected state was — what was agreed or assumed
- What needs to resolve — the specific change or outcome required
Rules:
- Keep it factual. No blame language, no emotional framing.
- Use the client's words where possible.
- One paragraph, 3-5 sentences maximum.
This brief is your anchor. If the conversation drifts, this is what you come back to.
Step 3: Situation Diagnosis (Difficult Conversation Prep — Condensed)
Review the issue brief and relationship context. Answer these four questions in order:
- What happened? The factual trigger — what changed, what was said, what failed, or what's been avoided. No interpretation yet.
- Why does it matter? The impact on the engagement, the deliverables, the relationship, or the client's outcomes.
- Why now? What makes this conversation necessary today rather than next week. If there's no urgency trigger, name the cost of continued delay.
- What's the real issue? Often the surface problem masks a deeper pattern. A scope creep conversation is really about the client not respecting your boundaries. A deliverable quality conversation is really about misaligned expectations set during onboarding. Name the deeper issue.
Pattern flag: If the user indicated this is a recurring issue, name the pattern explicitly. The script changes significantly for a pattern conversation vs. a one-time issue — patterns require naming the pattern itself, not just the latest instance.
Format: Short prose paragraphs. Bold each question as a label.
Step 4: Core Message Distillation
Write the uncomfortable truth in one sentence. Rules:
- No softening language ("I feel like," "it seems like," "I might be wrong but")
- No blame ("you didn't," "you failed to")
- States the situation as a fact and its impact as a consequence
- Pattern: "[What's happening] is [impact], and we need to [what changes]."
Then write three versions:
- Direct version — the clearest, most unvarnished statement
- Diplomatic version — same content, framed with partnership language
- Recommended version — which to use and a one-sentence explanation based on the relationship context
Step 5: Response Anticipation
Map the four most likely client reactions. For each, provide the actual words to say — not a description of what to do.
| Reaction | They Say | You Say |
|---|---|---|
| Defensiveness — "That's not what happened" / "We gave you everything you asked for" | [Likely words based on context] | [Acknowledge their perspective without conceding the point. Exact language.] |
| Deflection — "Let's focus on the bigger picture" / changes the subject | [Likely words based on context] | [Redirect back to the core issue. Exact language.] |
| Counter-accusation — "Well, your team was also late on..." | [Likely words based on context] | [Acknowledge what's valid, separate the issues. Exact language.] |
| Silence or withdrawal — Goes quiet, disengages, says "I need to think about it" | [Likely behavior based on context] | [Create space while maintaining the frame. Exact language.] |
Rule: Every cell in the "You Say" column must contain exact words the consultant can say out loud. "Acknowledge their concern" fails. "I hear you, and I understand that timeline felt tight on your end. Here's what I need us to address" succeeds.
Step 6: Build Conversation Script
Opening (first 90 seconds)
Write out the exact opening — the first 2-3 sentences the consultant will say:
- One sentence of context: "I want to talk about [topic] because [reason]."
- The core message (recommended version from Step 4).
- The desired outcome: "What I'd like to get to is [specific outcome]."
Rule: The consultant should be able to deliver this opening from memory after reading it twice. If it's too long or too complex, cut it down.
Bridge to Dialogue
One question that invites the client's perspective. Examples: "How are you seeing this?" or "What's your read on where we are?"
Rule: Do NOT open with the question. Deliver the message first, then ask.
If It Goes Well
- Confirm: The exact words to confirm the agreed path forward
- Next actions: Specific actions with owners
- Check-in: When to reconnect and verify follow-through
If It Goes Sideways
- Reset phrase: One sentence that re-centers the conversation. "I think we both want [shared goal]. Let's talk about how to get there."
- Boundary statement: What you will and won't accept going forward. Exact language.
- Graceful exit: How to end the conversation if it's not productive. "I can see this needs more thought. Let's reconvene [when] after we've both had time to sit with it."
Step 7: Scope Creep Response (If Applicable)
Gate: Only run this step if the user indicated scope is at issue. If scope is not at issue, skip to Step 8.
Using the SOW section and the out-of-scope work described, produce a scope response to introduce during or after the conversation.
7a. Request Analysis (Internal)
- Exact request: What the client is asking for, specifically
- Scope boundary: Where it sits relative to the original agreement (paraphrase the relevant scope language)
- Classification: Clearly out of scope / gray area / natural extension
- Client intent: Testing boundaries / genuinely confused / excited about a new idea
7b. Tone Calibration
Select based on analysis:
- Clearly out + strong relationship: Warm, direct, brief. "I'd love to do this — here's how we can make it work."
- Gray area + any relationship: Collaborative. "Let me clarify what's included so we're on the same page."
- Repeat boundary-testing + any relationship: Firm but professional. Name the pattern gently.
- New client + any request type: Extra warmth, extra clarity.
Never use: apologetic, defensive, legalistic, or passive-aggressive tone.
7c. Scope Response Draft
Build a 3-paragraph response (under 150 words):
Paragraph 1 — Acknowledge. Name the request specifically. Show genuine interest in the idea or need behind it. 1-2 sentences.
Paragraph 2 — Boundary. State what the current scope covers and where this request falls. "Our current engagement covers [X]. What you're describing — [specific request] — falls outside that scope." No apology. No hedging.
Paragraph 3 — Path Forward. Offer exactly one clear next step:
- "I can put together a quick change order for this — should take [timeframe] and would run [ballpark if appropriate]."
- "Let's add five minutes to our next call to talk through options."
- "This would be a great fit for Phase 2 — I'll add it to the backlog."
Close with: "Let me know how you'd like to proceed."
7d. Scope Response Quality Check (Internal)
| Check | Question |
|---|---|
| Specific | Does the response name the exact request and the exact scope boundary? |
| Warm | Would the client feel respected after reading this, not managed? |
| Structured | Is it acknowledgment → boundary → path forward, in that order? |
| Actionable | Is there exactly one clear next step? |
Run all four checks. Identify the weakest element. Rewrite it. Verify warmth wasn't sacrificed for clarity or vice versa. Present only the finished response.
Step 8: Outcome Anchoring
Define three outcomes, ranked:
- Best case — Client agrees, action plan set, relationship intact. State the specific agreement.
- Acceptable — Client hears the message, needs time, agrees to reconvene with a decision. State the specific commitment.
- Walk-away line — The minimum you need from this conversation for the engagement to continue. State it plainly.
If the conversation doesn't reach at least the acceptable outcome, document what that means for the engagement and what the next move is.
Step 9: Assemble Final Output
Present one unified document containing:
A. Issue Brief
The one-paragraph factual summary from Step 2.
B. Conversation Prep
# Difficult Conversation Prep: [Topic]
**Client:** [Name] | **Date:** [Date] | **Relationship tenure:** [Duration]
## The Core Message
**Direct:** [One sentence, no softening]
**Diplomatic:** [Same content, partnership frame]
**Recommended:** [Which to use and why]
## Anticipated Reactions
| Reaction | They Say | You Say |
|----------|----------|---------|
| Defensiveness | [Likely words] | [Your exact response] |
| Deflection | [Likely words] | [Your exact response] |
| Counter-accusation | [Likely words] | [Your exact response] |
| Silence | [Behavior] | [Your exact response] |
## Conversation Script
### Opening (90 seconds)
[Context sentence]
[Core message]
[Desired outcome]
### Bridge
[Question that invites their perspective]
### If It Goes Well
- Confirm: [What to say]
- Next actions: [Specifics]
- Check-in: [When]
### If It Goes Sideways
- Reset: "[Exact phrase]"
- Boundary: "[What you will/won't accept]"
- Exit: "[Graceful close]"
## Outcomes
1. **Best case:** [Specific outcome]
2. **Acceptable:** [Minimum viable outcome]
3. **Walk-away line:** [Non-negotiable minimum]
## Watch For
- Quick agreement without substance — probe with: "[Exact question]"
- [Pattern-specific signal based on situation]
C. Scope Response (if Step 7 was executed)
The finished scope response from Step 7c, ready to use.
D. Preparation Checklist
| Item | Status |
|---|---|
| Issue brief written (factual, no blame) | [complete / pending] |
| Core message distilled (three versions) | [complete / pending] |
| All four anticipated reactions scripted with exact language | [complete / pending] |
| Opening statement written (deliverable in 90 seconds) | [complete / pending] |
| Fallback position identified | [complete / pending] |
| Walk-away line defined | [complete / pending] |
| Scope response prepared (if applicable) | [complete / not needed] |
| Conversation scheduled within 48 hours | [pending] |
E. SOPs to Trigger
- [ ] Change Order Management — if the conversation results in a scope adjustment, run the change order SOP to document the change before any out-of-scope work begins
- [ ] Mid-Engagement Review — if this conversation surfaced issues that warrant a broader engagement check-in
Quality Check
| Check | Pass? |
|---|---|
| Core message states the uncomfortable truth in one sentence without hedging | |
| Every anticipated reaction has a specific, word-for-word response (not a description) | |
| Opening can be delivered from memory after reading it twice (not too long, not too complex) | |
| Three outcomes are specific enough to recognize in real-time during the conversation | |
| Tone is direct without being aggressive, warm without being weak | |
| Issue brief is factual — no blame language, no emotional framing | |
| Script includes at least one moment where the consultant invites the client's perspective | |
| Script does not exceed 5 minutes of talking — consultant should listen more than speak after the opening | |
| Fallback position is identified and the consultant knows their floor before entering | |
| If scope is at issue: response follows acknowledge → boundary → path forward structure | |
| If scope is at issue: response offers exactly one next step, not multiple options | |
| If scope is at issue: response is under 150 words | |
| Conversation is scheduled within 48 hours of completing this prep | |
| Pattern flagged if this is a recurring issue (script adjusted accordingly) |
Rules
- Script the opening, not the whole conversation. You can control the first 90 seconds. After that, you should be listening more than speaking. Don't over-script beyond the opening and response frameworks.
- Collect all inputs in one pass. Do not scatter prompts across multiple turns. Ask once, flag gaps, keep moving.
- Exact words, not descriptions. Every "You Say" entry must contain words the consultant can say out loud. "Acknowledge their concern" is not a script — it's a note to self that will evaporate under pressure.
- Never soften the core message to the point where it disappears. "I wanted to check in on how things are going" is not a difficult conversation — it's avoidance wearing a professional mask.
- Don't assume the worst about the client. The script should respect their intelligence and assume good faith until proven otherwise.
- Prep when you're clear-headed, not frustrated. Prep done while irritated produces language that sounds adversarial on the page and defensive in delivery. Run this SOP when you first identify the issue, not after it's been sitting for a week.
- Prepare for their reaction, not just your argument. If you haven't thought through the client's likely response, you'll get caught flat-footed and either over-concede or escalate unnecessarily.
- Schedule within 48 hours. Delayed difficult conversations get worse, not easier. The prep has a shelf life.
- Watch for quick agreement. Fast agreement to a difficult message often means they're agreeing to end the discomfort, not agreeing to the substance. Probe once: "I want to make sure we're aligned on specifically what changes. Can you tell me what you're hearing?"
- Flag patterns, not just incidents. If this is the third time the same issue has surfaced, the script must name the pattern — not just address the latest instance.
- One clear next step in the scope response. Never offer multiple options. One path forward only.
- Never apologize for scope boundaries. Boundaries are professional, not personal.
- Escape dollar signs as \$ for Notion compatibility.
- Flag inferred details. If a relationship assessment or reaction prediction was inferred rather than stated, mark it [INFERRED — verify].
Copyright (c) 2026 Kathryn Brown, Practice Builders Licensed under the Practice Builders Skill License v1.0 See https://practicebuilders.ai/license for terms.
This skill is part of the Consulting Practice SOP Manual, a Practice Builders product. Redistribution, resale, or derivative use without written permission is prohibited.