SLAs and escalation

Deadlines that act on their own.

A stage can carry an SLA window in hours. When the stage activates, the deadline is stamped on it, the inbox shows a countdown chip, and the request detail shows it on the stage map.

What happens on breach

A sweep runs every 30 minutes (and immediately on boot, to catch anything that happened while the service was down). For each breached stage it:

1. Marks the stage breached, which turns the chip red everywhere.
2. Writes SLA_BREACHED to the audit trail.
3. If the stage names an escalation role, reassigns every pending assignment to a holder of that role and records STAGE_ESCALATED.

Designing good SLAs

• Put short windows (4 to 24 hours) on gate stages where work queues up.
• Always pair an SLA with an escalation role on stages with a single approver. A deadline without an escape hatch is just a red badge.
• Escalation reroutes the work; it does not decide it. The escalation target still makes a real decision with the full context.