Skip to content
Back to Operations & SRE

Communications Protocol

Keep customers and stakeholders informed during incidents with a predictable cadence and status page.

Why

Silence during an outage is worse than the outage — customers assume you don't know, don't care, or are hiding something, and they tweet about it, which turns a 12-minute technical issue into a 24-hour PR crisis. A comms protocol that posts to the status page within 5 minutes of detection, updates every 15 minutes, and declares resolution explicitly turns the same outage into a demonstration of competence. The comms lead is a distinct role from the IC so that customer updates go out in parallel with debugging, not after it. The cadence matters more than the content: "still investigating, next update in 15m" is far better than silence, because it proves someone is paying attention.

When

For every SEV1 and any SEV2 with user-visible impact. The comms lead is designated at incident start.

How

1. Designate a comms lead distinct from the IC
   - Comms lead writes customer updates; IC feeds them facts; they don't compete for the same head
   - For SEV1, comms lead is paged in parallel with the debugger

2. Post the first update within 5 minutes of detection
   - "We're investigating issues with [service]. Follow this page for updates."
   - Don't wait for root cause; the update is "we know, we're on it"

3. Update on a fixed cadence, not when you feel like it
   - Every 15 minutes for SEV1, every 30 for SEV2, even if the update is "still investigating"
   - State the next update time: "next update by 14:30" — this proves the cadence is real

4. Use a public status page, not just internal chat
   - status.yourcompany.com with an RSS/email subscribe; customers check it before they tweet
   - Internal channel (#incidents) for the response; status page for the outside world

5. Declare resolution explicitly and post a follow-up
   - "The issue is resolved. We'll post a postmortem summary within 5 business days."
   - Don't let the status page drift to "investigating" forever; explicit close builds trust

Example

# status.yourcompany.com updates (SEV1, checkout down)
14:08  Investigating — We're investigating elevated error rates on checkout.
       Next update by 14:23.
14:23  Identified — A deploy caused checkout to fail for null-discount orders.
       Rollback in progress. Next update by 14:38.
14:38  Resolving — Rollback complete, error rate at 0%. Monitoring to confirm.
       Next update by 14:53.
14:53  Resolved — Checkout is operating normally. A postmortem will be posted
       within 5 business days at status.yourcompany.com/postmortems/2026-03-15.

# Internal #incidents channel (parallel, more technical):
14:08  comms-lead: "status page updated to Investigating. Next at 14:23."
14:15  IC: "root cause found, rollback decision made" → comms-lead drafts Identified update
14:23  comms-lead: "posted Identified. Next at 14:38."
# Comms lead never debugs; they translate IC's facts into customer-safe language.

Common Problems

  • ⚠No public status page — customers have nowhere to check, so they tweet or open support tickets, which overwhelms support during the exact moment they're least able to help.
  • ⚠Updates only when there's news — silence between updates reads as 'nobody is paying attention'; a fixed cadence with 'still investigating' is better than silence.
  • ⚠Comms lead is also the IC — customer updates wait until the IC surfaces from debugging, so the first update goes out 20 minutes after detection.
  • ⚠Status page never explicitly says 'resolved' — it drifts from 'investigating' to silence, leaving customers unsure if the issue is over.

Best Practices

  • ✓Designate a comms lead distinct from the IC; they post to a public status page within 5m of detection and update every 15m on a fixed cadence.
  • ✓Use a public status page with RSS/email subscribe; customers check it before they tweet, deflecting support load during the incident.
  • ✓State the next update time in every post ('next update by 14:30'); a fixed cadence with 'still investigating' beats silence every time.
  • ✓Declare resolution explicitly and promise a postmortem within 5 business days; don't let the status page drift from 'investigating' to silence.