Metodologia FinOps — Cloud Financial Operations

Financial Operations para cloud — visibilidad completa de costos, optimización de rate y usage, gobernanza financiera, forecasting, y unit economics. Transforma cloud cost de un gasto opaco en un driver de decisiones de negocio con responsabilidad distribuida por equipos. [EXPLICIT]

---

Principio Rector

1. Everyone takes ownership of their cloud usage — FinOps is a cultural shift, not just a tooling exercise. Engineers, finance, and leadership share accountability for cloud spend. Sin ownership distribuido, la optimización es un juego de whack-a-mole centralizado. [EXPLICIT]

2. Decisions are driven by business value, not just cost reduction — a $10K/month service that generates $100K revenue is well-spent. El objetivo no es "gastar menos" sino "gastar mejor". Cost efficiency sin business context es austeridad ciega. [EXPLICIT]

3. Cloud cost is a variable, manageable expense — real-time data enables real-time decisions. A diferencia de infraestructura on-premise (CAPEX fijo), cloud es OPEX variable. Esto es una ventaja competitiva si se gestiona activamente; un riesgo financiero si se ignora. [EXPLICIT]

---

Inputs

Parametros:
  MODO:         [piloto-auto | desatendido | supervisado | paso-a-paso]
  PROFUNDIDAD:  [express | standard | deep]
  FORMATO:      [ejecutivo | tecnico | hibrido]
  CLOUD:        [aws | azure | gcp | multi-cloud]
  ALCANCE:      [visibility | optimization | governance | full]

Modos de autonomia:
  - piloto-auto: Auto para analisis y recomendaciones, HITL para decisiones de compra (RI/SP). [EXPLICIT]
  - desatendido: Cero interrupciones. Analisis completo con supuestos documentados. [EXPLICIT]
  - supervisado: Autonomo con checkpoint en compras comprometidas y cambios de gobernanza. [EXPLICIT]
  - paso-a-paso: Confirma cada seccion, cada recomendacion, y cada calculo. [EXPLICIT]

Deteccion automatica:
  - Si existe billing data o Cost Explorer → CLOUD se infiere del provider
  - Si el input menciona "RI" o "savings plans" → ALCANCE incluye optimization
  - Si el input menciona "tags" o "allocation" → ALCANCE incluye visibility
  - Si el input menciona "forecast" o "budget" → ALCANCE incluye governance
  - Default: MODO=piloto-auto, PROFUNDIDAD=standard, ALCANCE=full, CLOUD=aws

---

When to Use

• Implementar programa FinOps desde cero (greenfield)
• Optimizar costos cloud existentes (rate optimization, right-sizing, idle detection)
• Diseñar estrategia de tagging y cost allocation
• Implementar showback/chargeback para equipos o business units
• Crear forecasting y budget management para cloud spend
• Calcular unit economics (cost per transaction, cost per user, cloud P&L)
• Evaluar madurez FinOps y definir roadmap de mejora (Crawl/Walk/Run)
• Detectar anomalias de costo y establecer alertas

When NOT to Use

• Auditar arquitectura AWS especifica para AI → metodologia-aws-architecture-audit
• Diseñar arquitectura cloud nueva → metodologia-aws-architecture-design
• Implementar infraestructura cloud → metodologia-aws-architecture-implementation
• Migrar workloads a cloud → metodologia-cloud-migration
• Estimar costos de un proyecto nuevo → metodologia-cost-estimation
• Analisis financiero horizontal (no cloud) → metodologia-analisis-horizontal-financiero

---

Delivery Structure: 6 Sections

S1: Cost Visibility & Allocation

Establece la base de visibilidad: saber quien gasta que, donde, y por que. [EXPLICIT]

Load references:
  Read ${CLAUDE_SKILL_DIR}/references/finops-unit-economics.md

Tagging Strategy:

• Mandatory tags: Environment, Team, Project, CostCenter, Application, Owner
• Optional tags: Compliance, DataClassification, ExpirationDate
• Tag enforcement: SCPs (AWS), Azure Policy, Org Policies (GCP)
• Tag compliance target: >95% de recursos taggeados

Cost Allocation:

• Shared cost distribution: proportional, fixed, even-split
• Amortized vs. unblended costs
• Custom allocation rules para shared services (networking, security, platform)

Dashboards:

• Daily cost trend por team/project/environment
• Top 10 cost drivers (servicios, cuentas, regiones)
• Tag compliance dashboard
• Anomaly detection alerts

Entregable: Tagging strategy document, cost allocation model, dashboard specifications.

S2: Rate Optimization

Reducir el precio unitario de los recursos cloud sin cambiar el consumo. [EXPLICIT]

Load references:
  Read ${CLAUDE_SKILL_DIR}/references/finops-optimization-playbook.md

Reserved Instances / Savings Plans:

• Coverage analysis: % de usage estable cubierto por commitments
• Target coverage: 70-80% de baseline steady-state
• RI vs. SP decision matrix (flexibility vs. discount depth)
• Term selection: 1yr vs. 3yr (risk vs. savings)
• Payment option: All Upfront > Partial Upfront > No Upfront

Spot / Preemptible:

• Workloads elegibles: batch, CI/CD, dev/test, stateless processing
• Interruption handling strategy
• Spot savings estimation

Enterprise Discounts:

• EDP/MACC negotiation leverage points
• Committed spend tiers
• Private pricing agreements

Entregable: Rate optimization plan con savings estimados por tecnica, timeline de implementacion.

S3: Usage Optimization

Reducir el consumo eliminando waste y ajustando el tamaño de los recursos. [EXPLICIT]

Load references:
  Read ${CLAUDE_SKILL_DIR}/references/finops-optimization-playbook.md

Right-Sizing:

• CPU utilization <40% sustained → downsizing candidate
• Memory utilization <30% sustained → downsizing candidate
• GPU utilization <50% sustained → evaluate CPU alternative
• Right-sizing cadence: mensual para compute, quarterly para databases

Idle Resource Detection:

• Unattached EBS volumes, unused Elastic IPs, idle load balancers
• Dev/test environments running 24/7 (target: 10h/day × 5 days = 30% savings)
• Idle endpoints (SageMaker, API Gateway con zero traffic)

Scheduling:

• Non-production environments: auto-stop nights/weekends
• Batch workloads: schedule durante off-peak pricing windows
• Expected savings: 60-70% en non-production compute

Storage Tiering:

• S3 lifecycle policies: Standard → IA → Glacier → Deep Archive
• EBS gp3 vs. gp2 migration (20% savings, better performance)
• Snapshot retention policies

Entregable: Waste inventory, right-sizing recommendations, scheduling plan, estimated savings.

S4: FinOps Operating Model

Estructura organizacional, procesos, y cadencia para sostener FinOps. [EXPLICIT]

Load references:
  Read ${CLAUDE_SKILL_DIR}/references/finops-operating-model.md

Team Structure:

• FinOps Lead / Cloud Economist (central)
• Finance Partner (central)
• Engineering Champions (embedded per team)
• Executive Sponsor (C-level)

Cadence:

• Daily: automated anomaly alerts, cost dashboards review
• Weekly: team-level cost review (15 min standup)
• Monthly: cross-team optimization review, RI/SP coverage check
• Quarterly: business review, forecast update, maturity assessment

Maturity Model — Crawl/Walk/Run:

Dimension	Crawl	Walk	Run
Visibility	Basic tagging, monthly reports	Full allocation, daily dashboards	Real-time, automated anomaly detection
Optimization	Ad-hoc right-sizing	Systematic RI/SP, scheduling	Autonomous optimization, ML-driven
Governance	Reactive budgets	Proactive forecasting	Predictive, business-aligned KPIs
Culture	Centralized FinOps team	Team-level champions	Engineering-owned, self-service

Entregable: FinOps operating model document, RACI, cadence calendar, maturity assessment.

S5: Forecasting & Budgeting

Predecir y controlar el gasto cloud con precision creciente. [EXPLICIT]

Forecasting Methods:

• Trend-based: historical extrapolation (± seasonal adjustment)
• Driver-based: cost = f(users, transactions, data volume)
• Bottom-up: team budgets aggregated
• Target accuracy: ±10% (Crawl), ±5% (Walk), ±3% (Run)

Budget Management:

• Budget = forecast + buffer (10-15%)
• Alert thresholds: 50%, 75%, 90%, 100%, 110%
• Escalation path per threshold
• Reforecast cadence: monthly (Crawl), bi-weekly (Walk), weekly (Run)

Anomaly Detection:

• Baseline: rolling 30-day average per service/account
• Alert trigger: >20% deviation from baseline
• Root cause analysis: new deployment, traffic spike, config change, pricing change
• Response SLA: investigate within 4 hours for >$1K anomalies

Variance Analysis:

• Budget vs. Actual vs. Forecast (monthly)
• Variance decomposition: rate change, usage change, scope change
• Trend analysis: MoM, QoQ growth rate

Entregable: Forecast model, budget framework, anomaly detection rules, variance report template.

S6: Unit Economics & Business Alignment

Conectar cloud cost con metricas de negocio para decisiones informadas. [EXPLICIT]

Load references:
  Read ${CLAUDE_SKILL_DIR}/references/finops-unit-economics.md

Unit Cost Metrics:

• Cost per transaction (CPT)
• Cost per active user (CPAU)
• Cost per API call
• Cost per GB processed/stored
• Infrastructure cost as % of revenue

Margin Analysis:

• Gross margin impact of cloud cost
• Cloud cost per product line
• Customer-level profitability including infra
• Break-even analysis per feature/service

Cloud P&L:

• Revenue attribution per cloud workload
• Direct vs. indirect cloud costs
• Shared services allocation
• Cloud ROI per business initiative

Showback/Chargeback:

• Showback: visibility sin billing impact (Crawl/Walk)
• Chargeback: actual cost transfer to business units (Run)
• Allocation model: direct attribution + proportional shared costs
• Billing cadence: monthly with T+5 reconciliation

Entregable: Unit economics dashboard, cloud P&L, showback/chargeback model, business alignment scorecard.

---

Trade-off Matrix

Profundidad	Scope	Depth	Effort	When to Use
Express	Visibility + quick wins	Top-line analysis	2-3 dias	Quick cost review, executive briefing
Standard	All 6 sections	Full analysis	5-8 dias	FinOps program launch, quarterly review
Deep	All 6 + implementation guides	Full + tooling setup	10-15 dias	Enterprise FinOps transformation

---

Assumptions

1. Acceso a billing data (Cost Explorer, billing console, o export)
2. Al menos 3 meses de historico de costos para analisis significativo
3. Tagging parcial existente (si no hay, S1 toma prioridad critica)
4. Stakeholders de engineering y finance disponibles
5. Cloud provider principal identificado (multi-cloud aumenta complejidad)

Limits

1. NO ejecuta compras de RI/SP — recomienda y el usuario decide/ejecuta
2. NO modifica infraestructura — identifica oportunidades (ver skills de implementation)
3. NO reemplaza Cost Explorer o herramientas nativas — las complementa con metodologia
4. NO provee consejo financiero regulado — es operational efficiency, no investment advice
5. NO audita seguridad ni compliance — solo cost implications (ver skills de audit)

---

Edge Cases

1. Startup sin historico: Usar industry benchmarks y driver-based forecasting. Establecer tagging desde dia 1. Focus en S1 y S4 antes de optimizar. [EXPLICIT]

2. Multi-cloud environment: Normalizar costos cross-provider. Usar herramientas multi-cloud (CloudHealth, Apptio, Spot.io). Allocation model debe ser provider-agnostic. [EXPLICIT]

3. Acquired company integration: Mapear tagging strategies, unificar cost centers, reconciliar contratos de RI/SP, consolidar billing accounts. [EXPLICIT]

4. Massive AI/ML spend: AI workloads tienen patrones unicos (GPU burst, training vs. inference). Separar AI cost center. Evaluar Inferentia/Trainium, spot para training. [EXPLICIT]

5. Regulated industry: Cost data puede contener metadata sensible. Chargeback reports deben cumplir data governance. Shared cost allocation debe ser auditable. [EXPLICIT]

Manejo de Inputs Ambiguos

• Si el cloud provider no se especifica: asumir AWS (default), confirmar si multi-cloud.
• Si el MODO no se especifica: usar piloto-auto (default).
• Si la PROFUNDIDAD no se indica: usar standard (default).
• Si el ALCANCE no se indica: usar full (default). Si el contexto menciona solo "costos" → visibility + optimization. Si menciona "FinOps program" → full.
• Si no hay billing data disponible: documentar como blocker, producir framework teorico basado en arquitectura conocida.
• Si el historico es <3 meses: advertir sobre precision limitada en forecasting, usar benchmarks de industria.

---

Validation Gate

El agente que ejecuta este skill debe verificar cada item antes de entregar el output al usuario.

• Tagging strategy definida con tags mandatorios y enforcement mechanism
• Cost allocation model documentado (direct, shared, amortized)
• Rate optimization analizado: RI/SP coverage, spot eligibility, enterprise discounts
• Usage optimization: right-sizing, idle detection, scheduling, storage tiering
• FinOps operating model: roles, RACI, cadence, maturity assessment
• Forecasting model seleccionado con target accuracy
• Budget framework con thresholds y escalation paths
• Anomaly detection rules definidas con response SLA
• Unit economics calculados (CPT, CPAU, cost as % revenue)
• Showback/chargeback model definido segun madurez
• Savings estimados cuantificados por categoria (rate, usage, waste)
• Roadmap priorizado con quick wins vs. strategic initiatives

---

Cross-References

Skill	Relacion
aws-architecture-design	Diseño AWS con cost considerations integradas
aws-architecture-audit	Auditoría que incluye cost optimization pillar
infrastructure-architecture	Infraestructura cloud donde se aplica FinOps
cloud-migration	Migracion cloud donde FinOps debe integrarse desde dia 1
cost-estimation	Estimacion de costos para proyectos nuevos
analisis-horizontal-financiero	Analisis financiero complementario (no cloud)

---

Output Format Protocol

if FORMATO == "ejecutivo":
  Executive dashboard: total spend, trend, savings opportunities, unit economics
  Audiencia: CFO, CTO, VP Engineering

if FORMATO == "tecnico":
  Full 6-section analysis + implementation details + tool configurations
  Audiencia: Cloud engineers, FinOps practitioners, DevOps

if FORMATO == "hibrido":
  Executive summary + technical deep-dive por seccion
  Audiencia: Engineering managers, technical leads reporting to C-level

Output Artifact

## {Organization Name} — FinOps Assessment & Roadmap

### Executive Summary
[Total cloud spend, trend, savings potential, maturity score, top recommendations]

### S1: Cost Visibility & Allocation [Tag Compliance: X%]
[Tagging strategy, allocation model, dashboard specs]

### S2: Rate Optimization [Potential Savings: $XX,XXX/month]
[RI/SP analysis, spot opportunities, enterprise discounts]

### S3: Usage Optimization [Waste Identified: $XX,XXX/month]
[Right-sizing, idle resources, scheduling, storage tiering]

### S4: FinOps Operating Model [Maturity: Crawl|Walk|Run]
[Team structure, RACI, cadence, maturity roadmap]

### S5: Forecasting & Budgeting [Forecast Accuracy: ±X%]
[Forecast model, budget framework, anomaly detection]

### S6: Unit Economics [CPT: $X.XX, Cloud/Revenue: X%]
[Unit costs, margin analysis, cloud P&L, showback/chargeback]

### Implementation Roadmap
[Phased plan with quick wins, medium-term, strategic initiatives]

### Appendix: Savings Detail
[Line-item savings by category with confidence level]

---

Fuente: FinOps Foundation. FinOps Framework (2024). | Storment, J.R. & Fuller, M. (2023). Cloud FinOps (2nd ed.). O'Reilly.