Multi-Audience Incident Comms Templates Prompt
Produce a coordinated set of incident communication templates tuned for three distinct audiences — internal responders, executives, and customers — so one source of truth fans out without contradicting itself.
- Target user
- Incident comms leads and communications managers
- Difficulty
- Intermediate
- Tools
- Claude, ChatGPT
The prompt
You are a communications lead who has run incident comms for major outages and knows the danger: three audiences, three needs, and the risk of saying three contradictory things from one set of facts. I will provide: - The incident's confirmed facts (impact, scope, start time, current status) - Audience constraints (regulatory, contractual SLAs, brand voice) - Channels available (status page, email, Slack, exec brief) - What is confirmed vs still under investigation Your job: 1. **Single source of truth** — define a fact block (impact, affected services, start time, current state, next update time) that every template draws from, so the three messages can never diverge on the facts. 2. **Internal / responder template** — terse, technical, action-oriented: current hypothesis, who owns what, what help is needed, where the bridge is. No spin. 3. **Executive template** — business framing: customer/revenue impact, severity, ETA confidence, decisions needed from leadership, regulatory exposure. Lead with the bottom line in two sentences. 4. **Customer template** — plain language, empathetic, no internal jargon or blame, no speculation. State impact, that you are on it, and the next-update time. Provide separate variants for status-page banner, in-app notice, and support-macro. 5. **Tone and what-not-to-say rules** — per audience, list forbidden phrases (no premature "fixed," no root-cause speculation to customers, no internal names externally). 6. **Update cadence** — set a per-audience rhythm and a rule that you always commit to a next-update time even when there is nothing new. 7. **Resolution and follow-up** — closing message per audience and a pointer to the eventual public postmortem if applicable. Output as: (a) the shared fact block, (b) three parallel template sets keyed to incident phases (detected, investigating, identified, monitoring, resolved), (c) a per-audience forbidden-phrase list, (d) the cadence schedule, (e) a worked example filling all three from one sample incident. Bias toward: factual over reassuring, committed next-update times, never speculating publicly, consistency across audiences.