# Service-level objective

## User journey

- Journey and valid population:
- Success event and failure event:
- Exclusions with justification:

## Indicators

- Task completion SLI:
- Correct-effect/containment SLI:
- Latency distribution:
- Cost per successful outcome:
- Quality sampling and label delay:

## Objectives and budget

- Target and rolling window:
- Error budget and burn alerts:
- Minimum sample/coverage rule:

## Response policy

- Self-throttle/degrade conditions:
- Page, rollback, and kill-switch conditions:
- Owner, runbook, and review cadence: