Design an incident comms playbook: audiences, cadences, and artifacts.
- User
Design an incident comms playbook: audiences, cadences, and artifacts.
Assistant
The user wants an "Incident Communications Playbook." This is a standard request for anyone in DevOps, SRE, or Incident Management.
What needs to be included?
- Objectives/Principles: Why communicate?
- Audiences: Who needs to know?
- Cadences/Frequency: When do we tell them?
- Artifacts/Templates: What do we send them?
- Roles & Responsibilities: Who does the talking?
- Escalation Paths: When do we move from one audience to another?
Deconstructing the core request:
-
Audiences:
- Internal Technical (Responders)
- Internal Stakeholders (Executives, Sales, CS)
- External Customers (Status page, Twitter, Support tickets)
-
Cadences:
- Initial Acknowledgment (Fast)
- Ongoing Updates (Regular intervals)
- Resolution (Final)
- Post-Incident (Deep dive)
-
Artifacts:
- Incident Ticket/Channel
- Status Page Updates
- Email blasts
- Post-Incident Review (PIR) / Root Cause Analysis (RCA)
Structure of the Playbook: I should organize this logically, perhaps by the lifecycle of an incident.
- Phase 1: Triage & Initial Comms
- Phase 2: Active Incident (Ongoing)
- Phase 3: Resolution
- Phase 4: Post-Incident
Drafting the Content - Section by Section:
-
Introduction: Keep it brief. Focus on transparency and accuracy over speed (speed is important, but accuracy prevents panic).
-
Audience Matrix (The "Who"):
- Responders: Need raw data, logs, commands.
- Executives: Need business impact (revenue, reputation) and ETA. No jargon.
- Customers: Need "Is it me?", "When will it be fixed?", "Am I safe?".
-
Cadence Strategy (The "When"):
- Rule of thumb: "Communicate every X minutes even if there is no update." (Silence breeds panic).
- SEV1: Every 15-30 mins.
- SEV2: Every 60 mins.
- SEV3: As needed/End of shift.
-
Artifacts (The "What"):
- The "Incident War Room" (Slack/Teams channel).
- Status Page.
- The RCA Document.
-
Templates (The "How"):
- I need to provide specific text templates. This is high value.
- Initial: "We are investigating..."
- Update: "Still investigating, identified root cause..."
- Resolved: "Service restored..."
-
Roles:
- Incident Commander (IC).
- Communications Lead (often the IC or a proxy).
Refining the Request: The prompt specifically asks for "Audiences, Cadences, and Artifacts." I will structure the response to highlight these three pillars explicitly, but weave them into a workflow so it's actionable.
Self-Correction/Critique during drafting:
- Initial thought: Just list the audiences. Correction: Map audiences to severity levels. Executives don't care about SEV3s usually.
- Initial thought: Provide one template. Correction: Provide a progression (Investigating -> Identified -> Monitoring -> Resolved).
Final Polish: Make it look like a document/playbook. Use formatting, bolding, and lists. Ensure the tone is professional and calm.
Plan:
- Core Principles: Establish the ground rules.
- The Framework (Table format): A matrix connecting Audience, Cadence, and Artifact.
- Deep Dive by Audience:
- Internal (Technical)
- Internal (Stakeholder)
- External (Customer/Public)
- Artifact Templates: Copy-pasteable examples.
- The Post-Mortem (The final artifact).
- Summary Checklist.
This looks like a solid, comprehensive guide. I will write it out now. This playbook outlines a structured approach to incident communication. The goal is to move from chaos to control by ensuring the right people get the right information at the right time.
1. Core Principles
Before defining the audiences and cadences, establish these ground rules for the Incident Commander (IC) and Comms Lead:
- Accuracy over Speed: Never speculate. If you don't know, say you are "still investigating."
- Single Source of Truth: All updates must funnel through one channel (e.g., the Status Page) to avoid conflicting narratives.
- Blamelessness: Focus on the system failure, not the human error.
2. The Audience Matrix
Different stakeholders require different levels of detail.
Audience Needs Tone Primary Channel 1. Responders (Engineering) Raw logs, timelines, technical next steps. Technical, imperative, shorthand. Slack/War Room channel, PagerDuty. 2. Executive Leadership Business impact, customer impact, ETA. High-level, strategic, "bottom line." Email, Executive Slack channel. 3. Customer Support (CS) Talking points to handle tickets, current status. Empathetic, scripted, FAQ-based. Slack (#incident-support), Internal Wiki. 4. Customers (External) Is it me? When will it be back? Is my data safe? Transparent, apologetic, jargon-free. Status Page, Twitter/X, In-App Banners.
3. Cadences (The Rhythm of Communication)
A. The "Heartbeat" (Active Incident)
You must establish a "comms heartbeat" so stakeholders aren't pinging you for updates.
- SEV-1 (Critical): Every 15–30 minutes.
- SEV-2 (High): Every 60 minutes.
- SEV-3 (Medium/Low): As needed or at resolution.
Rule: Even if there is no new information, you must send a "Still investigating, no update" ping at the cadence interval to prove the incident is active.
B. Lifecycle Triggers
Specific events trigger immediate comms outside the heartbeat:
- Acknowledgement: "We are aware of an issue."
- Identification: "We have identified the root cause."
- Mitigation: "The fix has been deployed; we are monitoring."
- Resolution: "The incident is resolved."
4. The Artifacts (Templates)
Artifact 1: The Internal "War Room" Update (Slack)
Purpose: To coordinate the technical response.
[STATUS: INVESTIGATING] Impact: Users cannot log in. RCA: Potentially related to Auth DB latency. Next Steps: @OnCall is restarting the auth cluster. ETA for next update: 15 mins.
Artifact 2: The Executive Update (Email/Slack)
Purpose: To prevent executives from surprising customers or the press.
Subject: [SEV-1] Update - Login Outage
Current Status: Active Incident. Business Impact: ~15% of users affected; login failures spiking. Customer Sentiment: High volume of support tickets (200+ in last hour). ETA: Expecting mitigation in 30 mins. Next Briefing: 2:00 PM EST.
Artifact 3: The External Status Page Update
Purpose: The "Public Record."
Investigating: We are currently experiencing a service interruption affecting user logins. Users may see a "503 Error" when attempting to sign in. Engineering is actively investigating the root cause. Next update in 30 minutes.
Identified: The issue has been traced to a database update. We are rolling back the change.
Resolved: The rollback is complete. Login services have been restored. We are monitoring for stability.
Artifact 4: The Post-Incident Review (PIR) / RCA
Purpose: Learning and closure (Sent 24–48 hours post-incident).
This document must include:
- Summary: What happened?
- Impact: How long? How many users? Revenue lost?
- Root Cause: The technical explanation.
- Resolution: How was it fixed?
- Action Items: "Prevention" tickets (Jira/Linear links) with owners.
5. Summary of Workflow
-
T-Minus 0 (Incident Starts):
- Create War Room (Artifact 1).
- Send Initial Status Page (Artifact 3) immediately (within 15 mins).
-
T-Plus 30 mins (Active):
- Send Heartbeat Update to Status Page.
- Send Executive Briefing (Artifact 2) if SEV-1.
-
Resolution:
- Update Status Page to "Resolved".
- Close War Room.
-
T-Plus 24-48 Hours:
- Publish PIR/RCA (Artifact 4).
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.