No. 25 · Nov 5, 2026 · Routing
Why critical incidents get missed
The gap is usually in routing, not response.
We put a lot of thought into our critical policy, from who gets called to how fast it escalates.
It's a little funny, because none of that helps if a critical incident never gets routed to it.
Say your billing API shipped last month, and at 3am its Postgres primary hits max_connections. Your critical policy should phone on-call right away, but nobody gets called.
That's because your routing rules predate the billing API, so none match its title. The incident falls to the default policy, which only posts to Slack.
This is hard to catch, because nothing looks broken. The incident gets triggered and routed, just to the wrong policy.
So maybe it's worth checking your default queue after every launch. If a critical incident is sitting there, it never got routed to your critical policy.