| name | workplace-conversation-prep |
| description | Prepares the user for a difficult workplace conversation by scripting the opening
statement, anticipated responses with branching dialogue, and de-escalation moves.
Covers feedback delivery, conflict resolution, boundary setting, and saying no to
scope creep. Produces a conversation script with branches.
Use when the user needs to have a difficult conversation at work, deliver tough
feedback, resolve a conflict with a colleague, set boundaries, or push back on
unreasonable requests.
Do NOT use for salary negotiation (use salary-negotiation-script), performance review
preparation (use performance-review-prep), or mentorship outreach (use mentorship-outreach).
|
| license | Apache-2.0 |
| metadata | {"author":"foundry-skills","version":"1.0.0","tags":"career writing template","category":"career-development","subcategory":"career-growth","depends":"","disclaimer":"none","difficulty":"intermediate"} |
Workplace Conversation Prep
When to Use
Use this skill when the user needs to prepare for a specific, upcoming difficult conversation at work -- one where the stakes are high enough that improvising would risk the relationship, the outcome, or the user's standing.
Trigger scenarios -- use this skill when:
- The user needs to deliver constructive feedback to a peer or direct report about a behavioral pattern (repeated lateness, poor communication, quality issues, missing commitments) and wants to do so without damaging the working relationship
- The user is experiencing ongoing conflict or tension with a colleague and needs to address it directly -- including passive-aggressive behavior, credit-stealing, exclusion from decisions, or interpersonal friction that has accumulated over multiple incidents
- The user needs to set a boundary with a manager or teammate -- around after-hours availability, scope additions, workload imbalance, or how they are spoken to -- and wants language that is firm without being adversarial
- The user needs to say no to a request, push back on an expanded scope, or flag capacity constraints to leadership, and must do so in a way that maintains credibility and relationships
- The user needs to hold a colleague accountable for a pattern of missed commitments -- deliverables, response times, attendance at critical meetings -- that has had a measurable impact on the user's work or the team's outcomes
- The user needs to address a sensitive personal workplace issue -- hygiene, inappropriate comments, social exclusion, undermining behavior -- where the risk of embarrassment or blowback is high
- The user is preparing upward feedback: sharing concerns with their own manager about behavior, decisions, or leadership style that is affecting the user's performance or wellbeing
- The user asks "how do I bring this up" or "what do I say when" about any sensitive work topic where social stakes, power dynamics, or emotional volatility are present
Do NOT use this skill when:
- The user is preparing to negotiate salary, compensation, or a promotion offer -- use
salary-negotiation-script, which addresses BATNA, anchoring, and compensation research frameworks
- The user is preparing for a formal performance review conversation where HR documentation, rating calibration, or improvement plans are involved -- use
performance-review-prep
- The user wants to reach out to a mentor or build a new professional relationship -- use
mentorship-outreach
- The conversation is primarily a project status update, alignment meeting, or stakeholder briefing -- this is not a difficult conversation, it is a communication task; use a meeting preparation framework instead
- The user is dealing with a situation that may constitute illegal harassment, discrimination, or retaliation -- this skill does not substitute for HR process, legal consultation, or formal complaint mechanisms; flag this directly and do not script a direct conversation as the primary path
Process
Step 1: Gather the Full Situational Context
Before writing a single line of script, collect enough information to produce a script that is specific, not generic. Generic scripts fail because the other person can feel the impersonality. Ask the user the following -- all of these matter:
- What happened, specifically? Get concrete incidents, not impressions. "Alex keeps missing deadlines" is insufficient. "Alex has submitted deliverables late in 5 of the last 7 sprints, each by 1-3 days, causing me to work overtime on 3 occasions" is actionable. Push for dates, frequencies, and documented impacts.
- What is the user's relationship to the other person? Direct report, peer, skip-level report, manager, senior leader, cross-functional partner, external vendor. The power dynamic fundamentally changes the opening language, the assertiveness level, and the acceptable escalation path.
- What is the desired outcome -- the minimum acceptable, the ideal, and the absolute non-negotiable? A conversation about deadline slippage might have: minimum = advance notice of delays, ideal = realistic sprint estimates agreed upon together, non-negotiable = the user stops working weekends because of others' misses.
- What has already been tried? Has this been raised before -- directly or indirectly? What happened? Repetition signals that normal approaches are not working and the script needs an escalation track.
- What does the user know about the other person's communication style, stress level, and likely reaction? A defensive person requires a different opening than a conflict-avoidant person or a high-emotional-reactivity person.
- What are the stakes if this conversation goes badly? If the answer is "we'd never work together again," the script needs more careful relationship preservation moves. If the answer is "I'd have to escalate to HR," the script needs documentation and a follow-up paper trail.
- Where and when will this happen? In-person vs. video vs. written. Private 1:1 vs. after a meeting. Impromptu vs. scheduled. Each context demands different pacing and staging.
If the user cannot answer some of these questions, help them identify the gaps -- and adjust the script to handle uncertainty by building in more checking and validation language.
Step 2: Identify the Conversation Type and Subtype
Not all difficult conversations have the same architecture. Identify the primary type, then the subtype, because each has a distinct opening move and structural logic.
Type 1: Feedback Delivery
- Subtype A -- Single incident: A specific, recent event that must be addressed before it becomes a pattern
- Subtype B -- Behavioral pattern: A recurring dynamic that has been observed multiple times
- Subtype C -- Upward feedback: The user is sharing concerns with someone senior to them
Type 2: Conflict Resolution
- Subtype A -- Active disagreement: A known, live dispute where both parties are aware of the tension
- Subtype B -- Simmering tension: Unspoken friction neither party has named, which is now affecting work quality
- Subtype C -- Values or working style clash: Differences in how each party approaches work that are causing friction
Type 3: Boundary Setting
- Subtype A -- Workload/scope boundaries: Pushing back on tasks being added without capacity conversation
- Subtype B -- Availability boundaries: Setting limits on after-hours contact, response time expectations
- Subtype C -- Behavioral boundaries: Addressing how the other person speaks to or treats the user
Type 4: Accountability
- Subtype A -- First conversation: First time addressing a specific pattern
- Subtype B -- Follow-up conversation: Pattern has been raised before with no improvement
- Subtype C -- Escalation-adjacent: One more incident will require formal escalation; the user is issuing a final informal warning
Type 5: Saying No / Pushing Back
- Subtype A -- Saying no to new scope: Declining an additional assignment while explaining capacity
- Subtype B -- Pushing back on a decision: Disagreeing with a direction while respecting authority
- Subtype C -- Flagging unsustainability: Communicating that current expectations are not maintainable
Identifying the correct type and subtype determines which opening framework to use (see Step 3) and which anticipated response branches to prioritize.
Step 3: Script the Opening Statement Using the Correct Framework
The opening statement is the highest-leverage part of the script. If the opening lands wrong, the rest of the conversation is damage control. The opening must accomplish three things in under 45 words of spoken language: signal intent, establish safety, and name the topic. Never cold-open into the substance.
Framework 1: Situation-Behavior-Impact (SBI) -- for Feedback Delivery
Structure: "In [situation], when you [specific observable behavior], the impact was [concrete effect on work, team, or user]."
- Situation = the specific context (not a generalization): "During last Tuesday's client call" not "always during client calls"
- Behavior = observable, factual, non-interpretive: "you cut off the client twice" not "you were rude"
- Impact = concrete and real: "the client emailed me afterward to ask if everything was okay with the team"
- SBI separates observation from judgment, which keeps the other person from immediately defending their character
Framework 2: Shared Interest Opening -- for Conflict Resolution
Structure: "We both care about [shared goal]. I want to talk about [dynamic that is getting in the way of it] so we can figure this out together."
- The shared interest must be real and specific, not generic ("we both want the project to succeed")
- Naming a shared goal before naming the problem reduces perceived threat
- The phrase "figure this out together" signals partnership, not prosecution
Framework 3: Constraint-First Opening -- for Boundary Setting
Structure: "I want to set some expectations about [topic] so we're both working from the same understanding."
- "Expectations" is less charged than "limits" or "rules" in most workplace cultures
- This opening does not require the other person to have done anything wrong -- it is framed as proactive alignment, not reaction to a violation
- For behavioral boundary violations (e.g., being talked over in meetings): "There's something I'd like to address with you privately because I want us to have a good working relationship."
Framework 4: Understanding-First Opening -- for Saying No / Pushing Back
Structure: "I know [acknowledge the legitimate importance of the request]. I want to be honest with you about [the capacity or scope constraint] so we can figure out the best way forward."
- Acknowledging importance before naming the constraint prevents the no from feeling dismissive
- "Be honest with you" signals transparency and respect, not obstruction
- Never start with the word "No" -- start with acknowledgment, end with the constraint, offer a path
Framework 5: Impact-on-Work Opening -- for Upward Feedback
Structure: "I want to share something with you because it's affecting my ability to do my best work, and I believe you'd want to know."
- This framing removes judgment from the manager's character and anchors the feedback in outcomes
- "I believe you'd want to know" pre-empts defensiveness by invoking the manager's identity as a good leader
- Never lead with "your behavior" or "what you did" when giving upward feedback -- always lead with the downstream effect on the user's work
Operational rules for all openings:
- Maximum 45 words of spoken text -- time this at under 30 seconds of speaking pace
- Request permission when possible: "Do you have 10 minutes?" signals respect and ensures the other person is in a receiving state
- Never open with "I just wanted to..." -- the word "just" minimizes the conversation's importance and signals anxiety
- Never open with a compliment designed to soften the blow (the feedback sandwich setup) -- it signals manipulation and the other person will brace for the critique immediately
- The opening must name the topic without delivering the full substance -- you are requesting engagement, not issuing a verdict
Step 4: Anticipate Responses and Script Branches
For every conversation type, there are four universal response archetypes. Each requires a distinct response technique, not just a different phrase.
Branch A: Defensive / Pushback ("That's not accurate / I didn't do that / You're exaggerating")
- Response technique: The Acknowledge-Anchor-Advance method
- Acknowledge: validate that their perspective is real, not that their facts are correct ("I hear that it doesn't feel that way from your side")
- Anchor: return to the concrete observation you opened with, without escalating ("What I observed was [specific incident again], and I'm sharing the impact it had on me specifically")
- Advance: propose moving forward rather than litigating the past ("I'm less concerned about who's right about the past than what we can do differently going forward")
- Do not try to win the factual argument -- defensiveness cannot be debated away
- Secondary escalation if defensiveness increases: name the dynamic directly ("I notice we're stuck on whether this happened. I'm not trying to build a case against you. I'm trying to solve a problem that's affecting my work.")
Branch B: Emotional / Upset ("I can't believe you think that / This is really hurtful / I feel attacked")
- Response technique: The Validate-Stabilize-Continue method
- Validate: name the emotion without judgment ("I can see this is landing hard, and that makes sense")
- Stabilize: offer a genuine pause option ("If you'd like a few minutes, I'm absolutely okay with that")
- Continue: if the person wants to proceed, return to the purpose without re-litigating ("When you're ready, I'd really like us to figure out a path forward together")
- Never say "don't be upset" or "this isn't personal" -- both phrases invalidate and escalate
- If someone cries: go quiet, offer water or a moment, do not rush, do not apologize for raising the topic
Branch C: Deflection / Redirect ("What about when you... / The real problem is... / Management is to blame for this")
- Response technique: The Acknowledge-Bracket-Redirect method
- Acknowledge: validate that the deflected point may have merit ("That's a fair point and worth discussing")
- Bracket: explicitly defer the deflection ("I want to make sure we give that the attention it deserves -- separately")
- Redirect: return explicitly to the original topic ("For right now, can I ask that we focus on [original topic]? That's the most urgent piece for me.")
- Do this maximum twice. If the person deflects a third time, name the pattern: "I notice we keep moving away from [original topic]. I want to understand if there's something about it that feels difficult to address directly."
Branch D: Agreement / Acceptance ("You're right / I didn't realize / I'm sorry about that")
- Response technique: The Receive-Clarify-Commit method
- Receive: accept the acknowledgment without gushing ("I appreciate that, it means a lot that you're hearing me")
- Clarify: translate the agreement into a specific behavior change, because vague agreement evaporates ("Can I ask what that would look like going forward?")
- Commit: propose a concrete checkpoint ("Let's plan to check in on this in [2 weeks / after the next sprint / at our next 1:1]")
- Do not over-celebrate agreement -- it can feel patronizing and creates pressure on the other person
- Document the agreement in a brief written summary within 24 hours
Step 5: Build the De-Escalation Toolkit
De-escalation is a physical and linguistic practice. Script specific phrases AND identify the non-verbal moves that accompany them.
Linguistic de-escalation techniques (in order of escalation severity):
- Name the emotion without judgment: "I can see this is bringing up some strong feelings." -- Use at first signs of rising tension.
- Acknowledge-without-concede: "I understand why you see it that way, and I want to stay focused on what we can do about it." -- Use when the other person is asserting their position loudly.
- Name the dynamic: "I notice we're getting a little stuck. Can we take a breath and come at this differently?" -- Use when the conversation has circled twice without progress.
- Offer an explicit pause: "This conversation matters too much for us to rush it. Would 5 minutes help?" -- Use when emotional intensity is high enough that productive conversation is not possible.
- Invoke shared stakes: "We both have too much riding on this project to let this stay unresolved. I want us to come out of this with a plan." -- Use when the other person seems to be disengaging or stonewalling.
- Exit gracefully: "I can see we're not in the right place for this today. Let's schedule 30 minutes this week when we can both be fresh." -- Use as last resort before a conversation becomes damaging.
Physical de-escalation for the user:
- Slow the pace of speech -- anxiety speeds speech, which signals alarm to the other person
- Lower the vocal register slightly -- higher pitch under stress reads as aggression or panic
- Keep posture open: no crossed arms, lean slightly forward to signal engagement not confrontation
- If speaking by video: turn on camera, position at eye level, ensure neutral background to reduce environmental distraction
- Before the conversation: 4-7-8 breathing (inhale 4 counts, hold 7, exhale 8) reduces cortisol and steadies voice
Step 6: Define the Outcome Architecture
Every conversation needs three defined endpoints, because conversations do not always land where you intend.
Endpoint 1: Full resolution -- agreement reached
Script a closing that: (a) summarizes the specific agreement in one sentence, (b) names the concrete next step with a timeline, and (c) reaffirms the relationship with a genuine (not sycophantic) statement. Example architecture: "So we've agreed that [specific change]. I'll [user's commitment] and you'll [their commitment] by [date]. I'm glad we talked about this -- I'd rather deal with this stuff directly than let it build."
Endpoint 2: Partial resolution -- progress but not complete agreement
Script a closing that: (a) names what was agreed, (b) names what is still open, and (c) schedules a specific follow-up. Do not let "we need to think about it" be the outcome without a date attached. Example: "It sounds like we're aligned on [part A] and we're still figuring out [part B]. Can we set 20 minutes on Thursday to resolve the rest?"
Endpoint 3: No resolution -- follow-up required
Script a closing that: (a) thanks the person for the conversation without conceding the issue, (b) names the next step clearly, and (c) signals that the issue is not dropped. Example: "I appreciate you hearing me out. I'm going to think about what we've discussed and I'll follow up with a written summary of where we left things. I'd like to resolve this by [date]."
Documentation trigger: Any conversation that involves accountability, boundary violations, or potential future escalation should be followed by a written summary sent within 24 hours -- typically a Slack message or email with subject line "Summary of our conversation today." This is not punitive; it is protective and professional. Script this follow-up message as part of the deliverable.
Step 7: Conduct a Pre-Conversation Risk Assessment
Before finalizing the script, evaluate three risk factors that affect how the script should be calibrated:
Risk Factor 1: Power asymmetry
- Speaking upward (to a manager or senior leader): reduce directness by 15-20%, increase "impact on my work" framing, never name the behavior as a character issue, always include a solution offer
- Speaking laterally (peer): standard assertiveness, shared-interest framing
- Speaking downward (to a direct report): increase directness by 10%, include specific behavioral expectations, document more carefully for performance management continuity
Risk Factor 2: Prior relationship quality
- Strong relationship: can be more direct, less setup required, recovery from awkwardness is faster
- Neutral relationship: invest more in opening rapport before naming the issue
- Damaged or strained relationship: lead with the desire to repair ("I want us to work well together and that's why I'm bringing this up"), expect more defensiveness, pace the conversation slower
Risk Factor 3: Organizational culture
- High-feedback cultures (tech startups, consulting, performance-driven teams): direct feedback is expected and valued; script can be sharper
- Low-conflict cultures (traditional corporate, certain public sector, family-owned businesses): indirect opening is required, face-saving must be built into every response branch
- High-hierarchy cultures: never challenge authority directly; always frame as seeking guidance, not delivering judgment
Step 8: Assemble the Full Script and Rehearsal Notes
Compile all components into the Output Format template. Then add:
- A 30-second rehearsal note: the user should read the opening statement aloud at least twice before the conversation. Mouth-to-ear rehearsal catches phrasing that reads fine but sounds stilted when spoken.
- A trigger phrase list: 3-5 phrases the user should avoid saying, with replacements. For example: replace "I feel like you always..." with "In the last [specific timeframe], I've noticed..."
- A timing note: conversations of this type should be allocated 20-30 minutes and should not be scheduled as the last item before a hard deadline (lunch, end of day, before a major meeting), because if the conversation runs long or ends badly, neither party should have to immediately perform in a high-stakes context.
Output Format
Produce the following complete script document. Every section must contain real words, not placeholders or brackets -- if the user hasn't provided a name, use "your colleague" or "your manager." Every quoted block must be word-for-word usable language.
## Workplace Conversation Script: [Conversation Type] -- [Subtype]
**Context at a Glance**
| Field | Detail |
|-------|--------|
| Situation | [Specific description of what happened and why this conversation is needed] |
| Other person | [Name or role, relationship to user, power dynamic] |
| Communication style (if known) | [Defensive / emotional / analytical / indirect / direct] |
| Your goal -- minimum | [The least you need from this conversation for it to be worth having] |
| Your goal -- ideal | [What a fully successful outcome looks like] |
| Your non-negotiable | [The thing that cannot be conceded regardless of pushback] |
| Setting | [When, where, how long, in-person / video / written] |
| Prior attempts | [Has this been raised before? What happened?] |
---
### Pre-Conversation Checklist
- [ ] I have written down the 2-3 specific incidents I will reference (dates, observable behaviors, concrete impacts)
- [ ] I have rehearsed the opening statement out loud at least twice
- [ ] I have blocked 30 minutes and confirmed the other person is available and in the right headspace
- [ ] I know what my non-negotiable is and will not concede it under pressure
- [ ] I have a written summary template ready to send within 24 hours of the conversation
---
### Opening Statement
**Framework used:** [SBI / Shared Interest / Constraint-First / Understanding-First / Impact-on-Work]
> "[Exact spoken words -- under 45 words, ready to say aloud]"
**Why this opening works:** [2-3 sentences explaining the strategic framing choice for this specific situation]
**Phrases to avoid in the opening:**
| Avoid | Replace with |
|-------|-------------|
| "[Problematic phrase 1]" | "[Better alternative]" |
| "[Problematic phrase 2]" | "[Better alternative]" |
| "[Problematic phrase 3]" | "[Better alternative]" |
---
### Response Branch A: Defensive / Pushback
**What they are likely to say:**
> "[Most likely defensive response, as specific as possible given the context]"
**Your response (Acknowledge-Anchor-Advance):**
> "[Full spoken response -- acknowledge their perspective, anchor to your specific observation, advance toward resolution]"
**If they escalate further:**
> "[Escalated de-escalation response -- name the dynamic, offer path forward]"
**If they make a counter-accusation:**
> "[Response that acknowledges the accusation without abandoning the original topic]"
---
### Response Branch B: Emotional / Upset
**What they are likely to say or do:**
> "[Most likely emotional response -- anger, hurt, withdrawal, tears]"
**Your response (Validate-Stabilize-Continue):**
> "[Full spoken response -- validate the emotion, offer stability, invite continuation when ready]"
**If they need time:**
> "[Pause offer with specific follow-up commitment -- do not leave it open-ended]"
**What NOT to say:**
- Do not say: "[Common but damaging phrase]"
- Do not say: "[Second common but damaging phrase]"
---
### Response Branch C: Deflection / Redirect
**What they are likely to say:**
> "[Most likely deflection -- blame shift, whataboutism, subject change, minimizing]"
**Your response (Acknowledge-Bracket-Redirect):**
> "[Full spoken response -- acknowledge the deflected point, explicitly defer it, redirect to original topic]"
**If they deflect a second time:**
> "[Second redirect -- firmer, names the pattern without accusation]"
---
### Response Branch D: Agreement / Acceptance
**What they are likely to say:**
> "[Most likely acceptance response -- acknowledgment, apology, willingness to change]"
**Your response (Receive-Clarify-Commit):**
> "[Full spoken response -- receive the agreement, clarify into a specific behavior, propose a checkpoint]"
---
### De-Escalation Toolkit
| Situation | Technique | Exact Words to Use |
|-----------|-----------|-------------------|
| Conversation gets heated | Acknowledge-without-concede | "[Specific phrase]" |
| Person shuts down or goes silent | Validate and offer pause | "[Specific phrase]" |
| Person raises unrelated grievances | Acknowledge-Bracket-Redirect | "[Specific phrase]" |
| You feel yourself losing composure | Self-regulate and name it | "[Specific phrase]" |
| Conversation has circled twice with no progress | Name the dynamic | "[Specific phrase]" |
| Person becomes aggressive | Boundary statement + exit offer | "[Specific phrase]" |
---
### Closing the Conversation
**If full agreement is reached:**
> "[Closing statement: summary of agreement + specific next step with timeline + relationship reaffirmation]"
**If partial agreement is reached:**
> "[Closing statement: name what was agreed + name what is still open + schedule specific follow-up]"
**If no resolution is reached:**
> "[Closing statement: thank without conceding + name the next step + signal the issue is not dropped]"
---
### Post-Conversation Actions
- [ ] Send written summary within 24 hours (see template below)
- [ ] Calendar the agreed follow-up checkpoint
- [ ] If pattern continues after [specific timeframe], escalate to [next step -- manager, HR, skip-level]
- [ ] Note the date, content, and outcome of this conversation in a personal log
**Written Summary Template (send within 24 hours):**
Subject: Summary of our conversation today
"Hi [Name], I wanted to follow up on the conversation we had [today/this morning/this afternoon]. We discussed [topic], and I understood us to agree on [specific agreement]. My next step is [user's commitment] by [date], and I understood yours to be [their commitment] by [date]. Let me know if I've captured anything inaccurately. [Optional: I'm glad we talked about this.]"
---
### Rehearsal Notes
**Say this out loud before the conversation:**
> "[The opening statement one more time, formatted for reading aloud]"
**The one phrase to remember if things go sideways:**
> "[Single anchor phrase the user can fall back on when they lose their place]"
**Your non-negotiable (write it here, do not negotiate on this):**
> "[The specific outcome or behavior that is not up for compromise, stated in one clear sentence]"
Rules
-
Always produce word-for-word scripts, never advice about communication. The deliverable is exact language the user can speak aloud, not principles about "being clear" or "staying calm." Every quoted block must be ready to say verbatim. If the user hasn't given enough context to script precisely, ask the specific questions needed before producing the script.
-
Never use the feedback sandwich. Positive-negative-positive sequencing (compliment, criticism, compliment) is transparent, well-documented in psychological research to reduce the impact of the feedback, and signals manipulation. Use Situation-Behavior-Impact exclusively for feedback delivery. If the user asks to "soften it," add relationship-safety language to the SBI structure -- do not wrap it in false praise.
-
The opening statement must be under 45 words of spoken text. At a natural speaking pace of 130 words per minute, 45 words takes roughly 20 seconds. A longer opening is a monologue -- it gives the other person time to build a counter-argument before they have engaged. Short openings invite response. Test this by counting words, not by feel.
-
Anchor every claim in specifics, not patterns. "You always do X" triggers defensiveness and is factually unprovable. "In the last three sprints, this happened on five occasions: [dates]" is specific, verifiable, and harder to dismiss. Any script that includes the words "always," "never," "constantly," or "every time" must be rewritten to name specific incidents.
-
All I-statements must describe impact on the user, not judge the other person's intent. "You clearly don't care about the deadline" is an intent attribution and is almost always wrong. "When the deadline slips without notice, I end up working weekends -- that's the impact I'm trying to solve" describes the user's experience only. The other person cannot dispute the user's experience the way they can dispute an attribution of intent.
-
Every response branch must include a secondary response. Real conversations are non-linear. A single response per branch is not enough. Each branch must include at least one follow-up line: what to say if the first response doesn't work, if the person escalates, or if they go silent. Conversation scripts with only one move per branch collapse when reality departs from the script's assumptions.
-
Power dynamics must change the language, not just the tone. Speaking to a manager is not the same as speaking to a direct report with a softer voice. Specific vocabulary must change: upward feedback uses "I've noticed an impact on my work" not "your behavior"; downward feedback uses "I need X by Y" not "it would be great if you could"; lateral feedback uses "we both" and "I want to figure this out with you." Applying the wrong power register to a conversation can damage relationships and career standing.
Edge Cases
Edge Case 1: Upward Feedback -- Addressing a Manager's Behavior
This is the highest-stakes conversation type and requires the most careful construction. Key adjustments:
- Use exclusively Impact-on-Work framing: "When [situation], I find it harder to [work outcome]" -- never "when you [behavior], it makes me feel [emotion attributed to manager's fault]"
- Never frame the manager's behavior as a character flaw or a leadership deficiency -- frame it as information they may not have: "I wanted to share this because I believe you'd want to know how it's landing"
- Include an explicit "I may be missing context" clause: "I recognize you may have context I don't -- I wanted to share my experience so we're working from the same picture." This is not weakness -- it is political intelligence that reduces the manager's defensiveness without conceding the point
- Prepare for a power-play response (manager becomes dismissive, reminds user of hierarchy, minimizes the concern): script a graceful but firm response: "I understand, and I'm not trying to challenge your judgment. I wanted to make sure this information was available to you. I'll leave it with you."
- If the manager has previously exhibited retaliatory behavior: note this explicitly to the user and recommend a written record (journal entry noting the conversation date, content, and outcome) before the conversation happens, and potentially consulting HR beforehand about whether to share feedback through a formal channel instead
Edge Case 2: The Issue Has Been Raised Before With No Change
This is a follow-up accountability conversation, and it requires a materially different structure than a first conversation:
- Open by explicitly referencing the prior conversation: "We talked about this on [date] and agreed that [specific change]. I want to check in on where things stand."
- Do not re-litigate the original issue -- the prior conversation is documented fact. The current conversation is about follow-through.
- Introduce consequences in the script -- not threats, but honest statements of what will happen if the pattern continues: "I want to be transparent with you: if this continues, I'm going to need to involve [manager / team lead / HR] to figure out a path forward. I'd much rather resolve this between us."
- The language should be noticeably firmer than a first conversation. Softening language that was appropriate the first time now signals that there are no consequences for non-compliance.
- After this conversation, a written summary is not optional -- it is mandatory. If this ever reaches formal HR process, the written record of multiple conversations is essential documentation.
Edge Case 3: The User Finds the Other Person Intimidating or Has High Anxiety
Some users are cognitively and physiologically impaired by anxiety in confrontational conversations. Adjust the script and prep for this:
- Shorten the opening to the absolute minimum -- every extra word is an extra opportunity to lose the thread under stress. Aim for 20-25 words maximum.
- Create a single anchor sentence the user can return to if they lose their place: a one-sentence version of the entire purpose ("I'm here to figure out how we can solve [specific issue] together.")
- Include physical grounding recommendations: feet flat on the floor, hands on table or in lap, one slow breath before speaking. These are not soft suggestions -- they reduce cortisol levels measurably and stabilize vocal output.
- Recommend a "pre-game ritual": read the script twice, say the opening aloud once in a private space (bathroom, stairwell, car), and remind themselves of the non-negotiable outcome. Athletes call this routine. It works.
- Suggest a low-stakes first line to buy processing time: "Before I get into it -- how are you doing?" buys 20-30 seconds and normalizes the interaction before the stakes go up.
Edge Case 4: The Conversation Is About a Personal Topic
Personal topics (body odor, hygiene, inappropriate personal comments, social exclusion behavior) carry the highest risk of embarrassment and damaged relationship. Specific adjustments:
- Always request a genuinely private setting -- not a hallway, not after a meeting with others nearby. A private room with a closed door or a private video call.
- Lead with explicit relationship respect: "I want to share something with you privately because I respect you and I'd want someone to tell me if the situation were reversed."
- Keep the behavioral description factually minimal and non-graphic: "There's something about your personal presentation that I think may be affecting how you're landing with the team" rather than a detailed description that increases humiliation.
- Prepare for a shame response -- the person may shut down entirely, become extremely apologetic, or cry. Have a compassion-forward stabilization script ready: "I can imagine this is uncomfortable to hear, and I want you to know I'm sharing it because I want things to go well for you, not to embarrass you."
- Do not over-explain or repeat the observation. Say it once, clearly, briefly. A second repetition in the same conversation increases the shame spiral without improving the outcome.
- Consider whether the person might have an underlying medical condition (e.g., a metabolic condition causing body odor, a skin condition affecting presentation). The script should never diagnose but should leave graceful space: "I don't know what's behind it -- I just wanted you to have the information."
Edge Case 5: The Conversation is a No / Scope Pushback to a Senior Leader
Saying no upward -- especially to someone with authority over the user's projects, budget, or career -- requires a specific structure that is neither submissive nor adversarial:
- Never say no as a first word or first concept. Always open with acknowledgment of the importance and legitimacy of the request.
- Offer a trade-off instead of a refusal: "I can take this on if we de-prioritize [X] -- which would you prefer I focus on?" This reframes the conversation from refusal to resource allocation, which is a business problem rather than a compliance failure.
- Use data where possible: "I'm currently running at [X hours/week] on [existing commitments]. Adding this would require [Y hours], which would push [existing commitment] to [delayed date]. I want to make sure you have that trade-off before I commit."
- If the senior leader pushes back and insists: do not cave silently. Script a final assertive offer: "I want to make this work. If this takes priority, can we align on what comes off my plate to make room for it?" This puts the capacity conversation back where it belongs -- with the person assigning the work.
- Document the exchange in an email after the conversation: "Following up on our conversation today -- I wanted to confirm that [project X] is now the priority and that [project Y] will be delayed until [date] as a result. Please let me know if I've understood this correctly." This protects the user if the leader later claims they didn't know about the trade-off.
Edge Case 6: The Other Person Has a Very Different Communication Style or Cultural Background
Communication norms vary significantly across cultures, industries, and personality types. A script calibrated for a direct American tech culture will misfire in a high-context East Asian corporate environment, a German engineering culture, or a relationship-first Latin American business context. Adjust based on:
- High-context vs. low-context cultures: In high-context communication environments (many East Asian, Middle Eastern, Latin American, and Southern European cultures), indirect opening and relationship-building before the issue is far more important. Never lead with the problem in these contexts -- lead with the relationship, then circle to the issue gently and with more hedging.
- Analytical vs. emotional communication styles: Analytical communicators (common in engineering, finance, data-heavy environments) respond better to SBI with data: cite the specific instances, the measurable impact, and a proposed solution. Emotional communicators respond better to the shared-interest frame and more relational language.
- High-hierarchy cultures: In cultures where challenging authority is strongly taboo, upward feedback may need to be framed as a question rather than a statement: "I wanted to ask for your guidance -- I've been finding that when [situation], I'm struggling to [outcome]. How would you suggest I handle that?" This delivers the feedback without the frame of a subordinate issuing a critique.
Edge Case 7: The Conversation Has Already Gone Wrong and Needs Recovery
Sometimes the user comes to this skill after a conversation has already escalated, ended badly, or resulted in a damaged relationship. The script needed is not a preparation script but a recovery script:
- Open the recovery conversation by acknowledging the previous conversation explicitly: "I've been thinking about our conversation on [date] and I don't think it went the way either of us would have wanted. I'd like to try again."
- Do not re-open the original substantive issue immediately -- address the rupture in the relationship first.
- Script an acknowledgment of anything the user did that contributed to the bad outcome (without taking full responsibility for the other person's behavior): "I think I could have framed it differently, and I'd like to approach it more carefully this time."
- After the relationship repair, re-raise the original issue using a softer version of the same framework -- but do not abandon it. Recovery does not mean conceding the point.
- Timeline note: recovery conversations should not happen the same day or within 24 hours of the original escalation. Both parties need physiological and emotional reset time.
Example
User input: "I need to talk to my manager, Jordan, about the fact that I'm being excluded from meetings where decisions are being made about my own projects. Three times in the last month, I found out about a major scope change or decision after the fact -- once by accident in a Slack channel. Jordan is generally supportive but has a lot on their plate and I think they may not realize how much this is affecting my work. I want to be included in decisions on my own work without it sounding like a complaint. This is the first time I'm raising it."
Output:
Workplace Conversation Script: Upward Feedback -- Boundary Setting on Decision-Making Inclusion
Context at a Glance
| Field | Detail |
|---|
| Situation | User has been excluded from three major decisions about their own projects in the past month -- once learning about a scope change accidentally via Slack. Has not raised this before. |
| Other person | Jordan (manager), generally supportive, currently overloaded |
| Communication style | Supportive but stretched thin -- likely to be receptive but may be defensive if it sounds like an accusation of bad management |
| Your goal -- minimum | Jordan commits to including user in any decision that affects their project scope, timeline, or deliverables going forward |
| Your goal -- ideal | A standing agreement: user is on the invite list for all relevant project discussions, or is explicitly looped in before decisions are made |
| Your non-negotiable | User will not accept "I'll try to remember" as the outcome -- needs a structural solution, not a good-faith promise |
| Setting | 1:1 meeting, first available slot this week, 20 minutes, video or in-person |
| Prior attempts | None -- this is the first time raising it |
Pre-Conversation Checklist
Opening Statement
Framework used: Impact-on-Work (Upward Feedback) + Constraint-First (Boundary Setting)
"Jordan, there's something I want to flag because it's starting to affect my ability to deliver well on my projects, and I think it's something you'd want to know about. Do you have 10 minutes to talk through it?"
Why this opening works: This opening uses two key upward-feedback techniques simultaneously. "It's affecting my ability to deliver well" anchors the conversation in work outcomes rather than a critique of Jordan's behavior. "I think it's something you'd want to know about" invokes Jordan's identity as a good manager -- it pre-empts defensiveness by implying that Jordan's intent is aligned with the user's goals. Asking permission ("do you have 10 minutes?") ensures Jordan is in a receiving state rather than catching them off guard. The opening is 34 words -- well under the 45-word ceiling.
Phrases to avoid in the opening:
| Avoid | Replace with |
|---|
| "I feel like I'm being left out" | "I haven't been included in three project decisions in the past month" |
| "You keep making decisions without me" | "Decisions about my projects are being made without my input" |
| "I just wanted to mention..." | "I want to flag something that's affecting my work" |
Core Statement (Deliver After Opening Is Accepted)
"In the past month, there have been three times where a decision was made about one of my projects and I found out after the fact. The [Project A] timeline was shifted, the [Project B] scope changed, and last week I found out about a [Project C] decision through a Slack message -- not a direct conversation. Each time, I had to scramble to adjust my work without the context of why the decision was made. I'm not raising this as a complaint -- I think it may just be a process gap. But I wanted to make sure we figured out a way for me to be in the loop before decisions are finalized, not after."
Response Branch A: Defensive / Pushback
What Jordan is likely to say:
"I hadn't realized that was happening. Some of those conversations move fast -- I can't always loop everyone in in real time."
Your response (Acknowledge-Anchor-Advance):
"I completely understand -- I know things move quickly and there's no way you can be in every conversation. I'm not expecting to be consulted on every small thing. What makes a real difference is being included when the decision affects my scope, timeline, or deliverables directly. Could we put a standing rule in place: if a decision will change what I'm building or when I need to deliver it, I get a heads up before it's finalized? Even a quick Slack before the decision is made would give me what I need."
If Jordan escalates to dismissing the issue:
"I hear you, and I recognize you're juggling a lot. The reason I'm bringing it up is that each time this happens, I lose time recovering from the change instead of executing on it. It's not about process for its own sake -- it's about my ability to do good work for the team. Is there a way we can build something simple that prevents this from falling through the cracks?"
If Jordan implies the user should have been more proactive in attending meetings:
"That's a fair point, and I want to be in those conversations -- I just don't always know which ones to attend. Would it help if we identified a standing set of meeting types that I should be on by default? That way it doesn't fall on either of us to remember case by case."
Response Branch B: Emotional / Upset
What Jordan is likely to say or do:
"I'm really sorry -- I can't believe I've been doing that. I've been so overwhelmed I clearly haven't been thinking about how this is landing."
Your response (Validate-Stabilize-Continue):
"I appreciate you saying that, and I want you to know I'm not bringing this up to add to your plate. I'm bringing it up because I know you care about how the work goes, and I wanted to give us both a chance to fix it before it becomes a bigger problem. I think a small structural change would solve it completely."
If Jordan needs time:
"We don't have to solve the whole thing right now. If it helps to think on it and come back to me with a proposal, I'm completely fine scheduling 15 minutes later this week to lock in the specifics."
What NOT to say:
- Do not say: "Oh, it's not a big deal, don't worry about it" -- this concedes the point and the conversation was just for nothing
- Do not say: "I totally understand, you're so busy" -- this lets Jordan off the hook and signals that the behavior is acceptable
Response Branch C: Deflection / Redirect
What Jordan is likely to say:
"The real issue is that our team communication process is broken overall -- this isn't just about you, it's everyone."
Your response (Acknowledge-Bracket-Redirect):
"You might be right that it's a broader team communication issue -- and if that's the case, it's worth addressing for the whole team. I'd love to be part of that conversation. For right now though, can we solve the piece that's in front of me? Even if we fix the bigger process later, I'd like to leave today with an agreement about how project decisions get to me specifically."
If Jordan deflects a second time:
"I notice we keep landing on the bigger systemic question, which is a real one. I want to make sure I'm not leaving today without something concrete. Can we spend two minutes on the specific ask -- inclusion in decisions about my projects -- and then I'll flag the broader process question for our next team meeting?"
Response Branch D: Agreement / Acceptance
What Jordan is likely to say:
"You're absolutely right, this is something I should have caught. What would actually be helpful for you?"
Your response (Receive-Clarify-Commit):
"I really appreciate that, Jordan. The most useful thing would be: for any decision that changes my project's scope, timeline, or deliverables, I get a message or a quick invite before the decision is finalized -- not after. It doesn't have to be a formal meeting. A Slack DM with 'heads up, we're thinking about doing X' is all I need. Does that work? And can we plan to check in on this in a month to see if the pattern has changed?"
De-Escalation Toolkit
| Situation | Technique | Exact Words to Use |
|---|
| Jordan becomes defensive about being a bad manager | Acknowledge-without-concede | "I'm not saying this reflects badly on you -- I think it's a gap in the system, not a personal failure. I just wanted to bring it to your attention." |
| Jordan goes quiet or seems to shut down | Validate and offer pause | "I can see this is a lot to take in. We don't have to resolve everything right now -- I just wanted to surface it. Do you want to think on it and come back to me?" |
| Jordan raises other team members' issues as a redirect | Acknowledge-Bracket-Redirect | "Those are worth discussing too. For today, can we just focus on the project decision piece? I'll bring the rest to our next 1:1." |
| You feel yourself becoming apologetic or backing off | Self-regulate | [Take one breath] "I want to make sure I'm being clear -- this has had a real impact on my work, and I want us to solve it together." |
| The conversation circles without resolution | Name the dynamic | "I feel like we might be dancing around the solution. Let me try to make the ask really specific: [restate the single, concrete request]." |
| Jordan says "let's follow up" without committing | Secure the commitment | "I want to make sure we land on something concrete before we close out. Can we agree on [specific solution] and I'll send a summary after this?" |
Closing the Conversation
If full agreement is reached:
"This is really helpful, Jordan. To make sure I've got it right: going forward, for any decision that changes my project scope, timeline, or deliverables, you'll flag it to me before it's finalized -- and I'll make sure I'm attending [standing meeting] so I'm in the room. I'll send a quick summary after this so we both have it. I appreciate you hearing me out."
If partial agreement is reached:
"It sounds like we're aligned that I should be in the loop on project decisions. Let's finalize the exact mechanism -- standing invite, DM protocol, or something else -- by end of week. Can I send you a quick email tonight with two options so you can pick what works best for you? I'll follow up if I
