The bit nobody covers in these write-ups: small teams without dual-cloud failover budget. Last big GitHub outage cost me a deploy day. Not catastrophic but the kind of thing you don't budget for when GitHub is your single source of truth.

Status page is also still doing that thing where every component is green but in practice clone is hanging, push is timing out, actions are stuck. Per-service uptime is a managed number. The user-experience number is the one that matters and it's not in the post-mortem.