Incidents and status updates

When to post

Post when customers are affected and would otherwise be guessing. Tracking not updating, the API returning errors, logins failing. Not for a slow query nobody has noticed — a status page that cries wolf gets ignored on the day it matters.

While it is still unclear

Post anyway. "We are investigating reports that tracking updates are delayed. Next update in 30 minutes." is worth far more than a perfect explanation an hour later, and it stops the support queue filling with the same question.

Then keep the 30 minutes. A promised update that does not arrive is worse than not promising one.

What a good update says

• What is affected, in the customer's words — "container tracking is not refreshing", not "the ingest worker is backed up".
• What still works. Most incidents are partial, and people need to know what they can rely on.
• When you will say more. A time, not "shortly".

Afterwards

Resolve the incident on the status page — an unresolved banner outlives the outage and makes us look broken for days. If tickets came in during it, answer them individually: a status post is not a reply.