Skip to content
Daily AI Intel

AI Prompts for Businesses · Customer Service & Support Prompts

How can you prompt AI to write an apology email for a service outage or mistake

An effective outage-apology prompt asks for a direct, specific acknowledgment of what went wrong and its real impact, followed by what's being done about it — since vague apologies that avoid specifics tend to erode customer trust further rather than repair it.

Security disclaimer

This content is provided for defensive, educational purposes only. It is not a substitute for a qualified security assessment of your specific environment. Test any configuration change in a non-production environment first.

Key takeaways

  • Specific acknowledgment of what actually happened builds more trust than a vague, generic apology.
  • Stating concrete remediation steps matters more than extended expressions of regret.
  • Transparency about the cause (to the extent appropriate) is generally better received than vagueness.
  • Outage communications about security or data incidents need legal/compliance review before sending.

The Prompt

Write an apology email to customers about this incident: [describe what happened — outage, error, data issue, etc.]

Impact: [who/what was affected, and how — be specific]
Cause (if appropriate to share): [brief, honest explanation]
What we're doing about it: [concrete remediation steps or prevention measures]
Any compensation or make-good being offered: [if applicable]

Tone: direct and accountable, not defensive or overly formal. Acknowledge the specific impact before explaining anything.

Why This Prompt Works

Vague, heavily lawyered apology language (“we take this very seriously”) without concrete specifics tends to read as evasive and can damage trust further. Requiring specific impact detail and concrete remediation steps in the prompt produces a message that demonstrates genuine accountability rather than one that just performs regret.

How to Customize It

For incidents involving customer data or security, route the draft through legal and security teams before it goes anywhere near customers — what can and can’t be disclosed, and how, often carries real regulatory implications that go well beyond good communication practice.

Common Mistakes to Avoid

Don’t let an AI draft speculate about the cause of an incident before it’s actually been confirmed internally — an inaccurate public explanation of what went wrong can create more problems than the original incident, especially for anything involving security or data.

Bottom Line

The most trust-rebuilding outage apologies are specific about impact and concrete about remediation — and any incident involving data or security needs legal review before an AI-drafted version goes out, regardless of how good the draft reads.

Build a Better Prompt

Turn a simple form into a clear, ready-to-copy prompt with our free Prompt Builder — no blank page required.

Sources

  1. [1]Prompt engineering guide — OpenAI
  2. [2]CX Trends report — Zendesk
ET

Written by Editorial Team

Last updated August 3, 2026

Get one well-sourced answer a week

No spam. Unsubscribe anytime.