Cost Optimization

When to use

• Reduce measured cost while preserving required service quality.

Not for: performance for latency; architecture for unbuilt systems.

Steps

1. Establish comparable billing, workload and service-quality baselines; redact account details.

2. Attribute dominant controllable costs across actual compute, storage, network, services and model calls.

3. Compare savings, effort, reliability and recovery; apply the smallest authorized change.

4. Re-measure under comparable conditions and label observed, estimated and unmeasurable savings.

Output: Suggested paths, subject to human audit: COST.md only for a useful continuing ledger.

Verify

• Cost comparisons state period and workload; quality regressions and estimated savings are explicit.

References

• engineering-principles.md — shared decision boundaries, scope and verification.
• human-audit.md — document consent, readability and human review.
• references/cost-drivers.md — symptom→cause map for cloud spend: over-provisioning, idle resources, egress, storage lifecycle, reserved-capacity gaps, with per-platform levers