On-call & incident response

The right person, woken once.

Cascade takes an alert, works down your escalation policy until a human acknowledges it, and writes the timeline while everyone else is busy fixing things.

4Severity levels, your rules
30sEscalation granularity
24/7Support on every plan
0Per-alert charges
Escalation

What happens when something breaks

Pick a severity to see the path an alert takes. Every step is yours to change.

SEV1 — customers cannot use the product

Target acknowledgement: 3 minutes
  • Primary on-call pagedPush, then a phone call thirty seconds later. Do-not-disturb is overridden for SEV1 only.
  • Secondary on-call pagedIf nobody has acknowledged. The primary keeps getting called — it is not a handover.
  • Engineering lead pagedAnd an incident channel opens automatically with the alert history pinned.
  • Status page draft preparedWritten from the alert, waiting for one person to hit publish. Never auto-published.

SEV2 — degraded, workarounds exist

Target acknowledgement: 15 minutes
  • Primary on-call notifiedPush only. No phone call — this is not worth a 3am ring.
  • Phone call to primaryOnly if it is still unacknowledged after fifteen minutes.
  • Secondary on-call pagedPlus a note in the team channel so daytime staff can pick it up.

SEV3 — minor, one customer or one region

Target acknowledgement: next working hour
  • Queued to the team channelNo page. It waits for someone to claim it.
  • Assigned to the on-call rotaIf nobody claims it in an hour it lands on the on-call engineer's list for the shift.

SEV4 — informational

No acknowledgement expected
  • Logged, nobody pagedGrouped by cause and shown in the weekly digest. Useful for spotting a slow leak.
  • Noise reportWhich alerts fired most and never mattered, so you can delete them.
An engineering team working through an incident
The timeline

The post-mortem writes itself while you fix it

Every page, acknowledgement, deploy and status update lands on one timeline with timestamps. When the incident closes you have the first draft of the review already.

  • Alerts, acknowledgements and resolutions in one thread
  • Deploys and config changes pulled in from your CI
  • Chat messages you star get added to the timeline
  • Export as markdown, so the review lives in your repo
See the Product
Teams

What changed after switching

★★★★★

“Our alert noise dropped by two thirds in a month because the weekly report made it obvious what to delete.”

Platform lead · fintech, 40 engineers
★★★★★

“The escalation actually escalates. Before, an unacknowledged page just sat there until morning.”

SRE · logistics SaaS
★★★★★

“Post-mortems take an hour now instead of half a day of scrolling back through chat.”

Engineering manager · 12 engineers
Get started

Bring your worst alert

Thirty minutes, screen shared, using your real alerting. If it does not obviously help, we will say so.