COST-005 recommended Cost Governance
Cost governance
Cost attribution enables accountability. Regular reviews turn visibility into action.
Question to ask
"Which team is responsible for that $8k AWS charge?"
Pass criteria
- ✓ Costs attributable to projects/teams
- ✓ Tagging enforced (if used)
- ✓ Regular review cadence
- ✓ Reviews lead to actions
Fail criteria
- ✗ All costs in one bucket
- ✗ Tagging not enforced
- ✗ No regular reviews
- ✗ Reviews without actions
Verification guide
Severity: Recommended
Cost attribution enables accountability. Regular reviews turn visibility into action.
Check automatically:
# Check for cloud tagging policies
grep -riE "tag.*policy|required.*tag|cost.*tag|cost.*center|cost.*allocat" docs/ terraform/ --include="*.md" --include="*.tf" 2>/dev/null
# Look for tagging in infrastructure-as-code
grep -riE "tags.*=|labels.*=" terraform/ pulumi/ cdk/ --include="*.tf" --include="*.ts" --include="*.yaml" 2>/dev/null | head -20
# Check for cost review documentation
grep -riE "cost.*review|budget.*review|spend.*review|finops.*meeting" docs/ --include="*.md" 2>/dev/null
# Look for cost allocation documentation
grep -riE "cost.*attribution|cost.*allocation|chargeback|showback" docs/ --include="*.md" 2>/dev/null
Ask user:
- "Can you attribute costs to specific projects or teams?" (tags, separate accounts, manual)
- "Is tagging enforced for cloud resources?"
- "Do you review costs regularly? How often? Who participates?"
- "What actions come out of cost reviews?" (optimizations, cleanups, budget adjustments)
Cost attribution methods:
| Method | Pros | Cons |
|---|---|---|
| Resource tagging | Granular, flexible | Requires enforcement |
| Separate accounts | Clean separation | Overhead to manage |
| Manual allocation | Simple to start | Error-prone, labor-intensive |
| Usage-based split | Fair | Complex to implement |
Pass criteria:
- Costs attributable to projects/teams (tagging, accounts, or documented allocation)
- Tagging policy exists and is enforced (if using tags)
- Regular review cadence (monthly minimum)
- Reviews lead to actions (optimizations, cleanups, alerts)
Fail criteria:
- All costs in one bucket, no breakdown possible
- Tagging exists but not enforced (many untagged resources)
- Reviews are ad-hoc or never happen
- Reviews happen but no actions result ("we just look at the numbers")
Cross-reference with:
- COST-004 (need visibility to do governance)
- COST-003 (tool ownership enables accountability)
Evidence to capture:
- Cost attribution method (tags, accounts, manual)
- Tagging enforcement (policy exists? enforced?)
- Review cadence and participants
- Recent actions from cost reviews