Escalation Response Drafter

---
name: escalation-response-drafter
description: Use when drafting a formal response to a customer escalation, RCA, or executive complaint. Triggers on "escalation response", "RCA email", "respond to escalation", "exec escalation".
---

# Escalation Response Drafter

Escalations succeed or fail in the first response. A defensive response confirms every fear the customer had. A self-flagellating response signals weakness. The right response: honest acknowledgment, concrete remediation, specific owner and timeline, no more, no less.

## Required inputs

1. **What actually happened** — facts, not characterization
2. **What the customer said** in the escalation (verbatim if you have it)
3. **What was YOUR side's failure** — the honest read, not the legal-safe version
4. **What's fixable** — both the immediate issue and the systemic root cause
5. **Who's signing the response** (CSM rarely — usually VP CS, CEO, or product leader)
6. **What's already been done** since the issue surfaced

If the user gives you a sanitized version ("there was a service interruption"), push back: "What actually broke? How long? How was it communicated? The customer already knows — your response shouldn't be vaguer than their experience was."

## Structure

### 1. Subject + opening (no preamble)
- Specific subject: "Friday's outage and what we're doing"
- Opening line names the issue directly. No "thank you for bringing this to our attention."

### 2. Acknowledgment (1 short paragraph)
Own the failure without theatrical apology.
- BAD: "We are deeply sorry that you experienced an interruption..."
- GOOD: "We had a 3-hour outage Friday during your month-end close. Our incident comms were silent until the post-mortem. Both are on us."

### 3. What happened (factual, brief)
- Plain language, no jargon
- Root cause if known; "still in investigation" with date for the RCA if not
- Don't blame upstream providers — own it

### 4. Concrete remediation (the heart of the response)
For each issue, one bullet:
- **What we're doing** (specific action)
- **Who owns it** (named person, not a team)
- **By when** (specific date)
- **How you'll know it shipped** (proof point — a doc, a call, a metric)

Examples:
- "Engineering (lead: Sarah Chen, VP Eng) is shipping circuit-breaker improvements to the failed component by May 30. We'll send the post-deploy report."
- "I (Mark, your VP CS) am setting up a standing 15-min weekly check-in with you for the next 4 weeks, starting this Wednesday at 2pm ET."

### 5. Make-good (if appropriate)
A specific gesture, not vague credit.
- "We're crediting your account for $X covering the affected period."
- "We're extending your renewal deadline by 30 days to give you space to evaluate."
Don't offer make-good if it doesn't fit — sometimes the right response is the fix, not the gift.

### 6. Next step
One specific next conversation, with a calendar invite already sent (or a "I'll send a Calendly link" if you can't presume).
- "I've sent a calendar invite for Thursday 3pm ET. If that doesn't work, name a time and we'll move."

### 7. Sign-off
Title-bearing person on your side. Phone number if appropriate. No "warm regards" filler.

## Tone calibration

### For incident / outage escalations
Technical owner signs the RCA section. CS owns the relationship. Don't let comms team write it for you — customers smell PR-speak.

### For billing / contract escalations
Same structure, but "owner" should include Finance or Ops, not just CS.

### For relationship escalations (CSM not engaged, slow responses)
Hardest to draft. The remediation is "different ways of working," and you have to commit to specifics — new cadence, new owner, escalation path — not "we'll do better."

## Anti-patterns to strip

- "Thank you for your patience" — they don't want thanks, they want a fix
- "We take your feedback seriously" — empty signal
- "We're committed to your success" — every vendor says this
- Long apology paragraph followed by no plan — bigger insult than no apology
- Defensive language ("our SLA technically covers this") — true or not, it kills trust
- Passive voice ("mistakes were made", "the system experienced an issue") — own it actively

## Output format

A complete email, plus:
- Suggested CC list (AE? VP CS? CEO if escalation went to their CEO?)
- A 1-line note on whether the response should be paired with a phone call BEFORE sending — high-stakes escalations almost always should be.