Post a Public Status Page Update
The status page is the customer's ground truth. A vague component ("API"), a pinned "investigating" that never advances, or a silent edit all cost more trust than the outage itself.
Procedure
- Set the affected component as narrowly as the page allows — "Payments API", not "Platform" — so unaffected customers do not panic.
- Pick the level from impact:
Investigating->Identified->Monitoring->Resolved. Advance it as facts land; never leave it atInvestigatingonce you know the cause. - Write the body in customer terms: what they experience, whether data is safe, the workaround, and the next-update time in UTC.
- Post a NEW timestamped update rather than editing the old one; customers who screenshotted or were emailed the prior text must see the diff.
- Set the next-update time honestly and honour it; a status page that goes quiet reads as activity being hidden.
- Post the
Resolvedentry with the end time and a plain cause, even a brief one — an outage that ends with no closing note leaves customers waiting for the other shoe. - Keep internal incident slugs and stack names out of the public body.
- Mirror the entry into the internal incident channel so the two records agree.
- If the impact scope widens or narrows, post a new entry that says so rather than quietly changing the component.
Pitfalls
- Marking "Platform" degraded when one payment method failed, alarming everyone.
- Editing the previous update in place, so the page shows a single tidy message and no history.
- Marking
Resolvedbefore sustained baseline, then reopening ten minutes later. - Copy-pasting the internal update, leaking service names and internal ticket ids to customers.
- Leaving the page at
Investigatingfor hours after mitigation because no one owns the next status change. - Diverging from the internal channel, so support tells customers one story and the page shows another.
Verification
grep -E '^(Investigating|Identified|Monitoring|Resolved) [0-9:]+ UTC' incident/public-status.log
# passes when the component, status, and a next-update time exist for each entry and none is a silent edit
Related: keep the internal and public wording aligned by drafting with draft-a-customer-facing-incident-message.