Cloud Economist

Cloud Economist

When to Use

• Build TCO/NPV for cloud migration, repatriation, or hybrid
• Compare economic options — serverless vs containers vs VMs, managed vs self-run, regions
• Model commitment portfolio — coverage mix, term, break-even, flexibility value
• Analyze marginal cost of scale (next 10k users, new region, new product line)
• Quantify elasticity — cost at p50 vs p99 load; burst vs steady-state economics
• Prepare EA or commit negotiation inputs (volume curves, alternative structures)
• Run sensitivity analysis — usage growth, price changes, discount tiers, FX
• Link cloud spend to gross margin and unit economics for product/finance
• Evaluate multi-cloud cost of redundancy vs lock-in risk (economic framing)
• Support cloud-architect and enterprise-cloud-architect ADRs with cost models

When NOT to Use

• CUR analysis, idle resource cleanup, monthly budgets → finops-analyst
• GL mapping, amortization, fixed assets → compute-accounting-manager
• VPC design, landing zones, migration waves (technical) → cloud-architect
• Org CCoE and EA program design → enterprise-cloud-architect
• Cloud program portfolio, migration funding, executive EA sign-off → vp-of-cloud
• Issue trees and operating model (non-cloud-specific) → business-consultant
• Rack/GPU supply TCO in data centers → data-center-compute-supply-efficiency
• Implement infra or purchase commits → cloud-engineer, finops-analyst
• Deal-level sizing, RFP cost framing, build-vs-buy for a customer → solutions-architect

Related skills

Core Workflows

1. Scope and economic lens

Boundaries vs FinOps, architect, finance.

See `references/cloud_economist_scope.md`.

2. TCO and build vs buy

Cloud vs on-prem vs SaaS; NPV framework.

See `references/tco_build_vs_buy.md`.

3. Commitment portfolio economics

RI/SP/CUD/EA economic trade-offs.

See `references/commitment_portfolio_economics.md`.

4. Marginal cost and elasticity

Scale economics and demand shapes.

See `references/marginal_cost_and_elasticity.md`.

5. Scenario and sensitivity modeling

What-if and risk ranges.

See `references/scenario_sensitivity_modeling.md`.

6. Executive economic briefings

CFO and board narratives.

See `references/executive_economic_briefings.md`.

Outputs

• Economic model — assumptions, 3–5 year cash flows, NPV/IRR where applicable
• Option comparison table — cost, flexibility, risk, break-even
• Commitment recommendation memo — coverage, term, $ at risk, exit cost
• Sensitivity chart — drivers ranked by impact on spend or margin
• One-pager for ADR — recommended option with economic rationale
• Negotiation brief — volume forecast, walk-away, alternative structures

Principles

• Explicit assumptions — growth, utilization, discount, labor, egress
• Compare options at equal SLO — do not trade reliability without pricing risk
• Time value matters — NPV for multi-year commits and migrations
• Flexibility has value — price the option to scale down or pivot
• Hand off execution — economist models; finops-analyst operationalizes

Cloud economist scope

Table of contents

1. Role boundary 2. Questions economists answer 3. Inputs and outputs 4. Handoffs

Role boundary

Economist models; does not own billing consoles or terraform.

Questions economists answer

• Should we migrate workload X to cloud vs extend colo?
• What is NPV of a 3-year all-upfront commit vs pay-as-you-go?
• Which region minimizes TCO given latency constraints?
• How does gross margin change if egress doubles?
• What commit coverage maximizes risk-adjusted savings?
• Is multi-cloud worth +$Y/year for portability?

Inputs and outputs

Inputs:

• Usage forecasts (from product, FinOps history, architecture)
• List prices and discount schedules (EA, private offers)
• Labor rates (run vs build), migration one-time costs
• SLO and residency constraints from architecture

Outputs:

• Spreadsheet or notebook model with documented assumptions
• Executive summary with recommendation and risks
• Feeds ADRs, board decks, negotiation targets

Handoffs

Commitment portfolio economics

Table of contents

1. Instrument comparison 2. Coverage and term 3. Break-even math 4. Portfolio view 5. EA negotiation inputs

Instrument comparison

Price option value of flexibility when usage is volatile.

Coverage and term

Recommendations:

• Cover steady baseline (e.g., 60–80% of min monthly usage), not peak
• Shorter term when product/market uncertain; longer when stable 2+ years
• Mix instruments — do not put 100% in one SKU family
• Reconcile with finops-analyst utilization reports monthly

Break-even math

Simplified:

monthly_savings = on_demand_cost × discount_rate
payback_months = upfront_payment / monthly_savings   (if applicable)

Include:

• Amortization of upfront vs monthly commit fee
• Exchange/sell fees if provider allows
• Opportunity cost of prepaid cash (discount rate)

If break-even > expected stable life of workload, reject commit.

Portfolio view

At org level:

• Aggregate by payer account, region, instance family
• Avoid double coverage (RI + SP on same usage)
• Track utilization % of commit pool; target >85% on steady portfolio
• Model drawdown of EA commit balance vs on-demand overflow

EA negotiation inputs

Provide finance and enterprise-cloud-architect:

• Forecast spend by year (low/base/high)
• Mix of services (compute vs data vs SaaS marketplace)
• Walk-away — discount at which on-demand + spot is acceptable
• True-up risk if under-commit
• CPI/price increase clauses sensitivity

Economist does not sign contracts — supplies scenarios.

Executive economic briefings

Table of contents

1. One-pager structure 2. ADR economic appendix 3. CFO questions 4. Common pitfalls

One-pager structure

1. Decision — what we are choosing (one sentence) 2. Recommendation — option B; NPV or 5y TCO delta vs A 3. Why now — trigger (migration, EA renewal, margin pressure) 4. $ impact — year 1 run-rate and 5y cumulative (range) 5. Risks — usage miss, lock-in, egress surprise, labor 6. Next steps — owners, dates, FinOps/eng actions

Use charts: cumulative TCO lines, tornado for top 3 drivers.

ADR economic appendix

For cloud-architect ADRs attach:

Keep technical ADR body with architect; economist appendix is optional attachment.

CFO questions

Prepare answers:

• What is run-rate cloud COGS as % revenue?
• What happens to margin at 2× scale?
• How much is committed vs flexible next 12 months?
• What is true-up exposure on EA?
• What is cash impact of upfront vs monthly commit?
• How do we reconcile model to GL? (compute-accounting-manager)

Common pitfalls

• Sunk cost fallacy — past migration spend should not justify bad forward option
• Ignoring labor — managed service savings need net of reduced FTE or redeployment
• Peak pricing — modeling at max scale without probability
• Discount double-count — EA discount plus RI on same usage
• Egress omission — largest error in data-heavy migrations

Pair narrative with business-consultant only when decision is broader than cloud economics.

Marginal cost and elasticity

Table of contents

1. Marginal cost 2. Elastic workloads 3. Unit economics link 4. Egress and data economics

Marginal cost

Marginal cost = incremental cloud spend for one more unit of business output.

Examples:

• Cost per additional 1k MAU (compute + DB + bandwidth)
• Cost per new microservice at steady traffic
• Cost per TB stored per month

Calculate from:

marginal = (total_cost_at_Q2 - total_cost_at_Q1) / (Q2 - Q1)

Use tagged workloads where possible; allocate shared platform separately.

Elastic workloads

Plot cost vs utilization:

Peak-to-average ratio drives commit suitability — high ratio → less RI, more on-demand/spot.

Unit economics link

Bridge to product and finance:

gross_margin = revenue_per_unit - (cloud_cogs_per_unit + allocated_platform)

Cloud economist supplies cloud_cogs assumptions; finops-analyst tracks actuals; compute-accounting-manager maps to GL.

Flag when cloud COGS % threatens target margin at roadmap scale.

Egress and data economics

Often dominant marginal cost:

• Cross-AZ replication
• Internet egress
• Cross-region DR
• SaaS analytics export

Model $/GB by path; compare architecture alternatives (CDN, compression, regional aggregation).

Partner with cloud-architect for data flow changes.

Scenario and sensitivity modeling

Table of contents

1. Scenario set 2. Key drivers 3. Sensitivity tornado 4. Model hygiene

Scenario set

Standard three scenarios:

Add stress for regulatory or outage-driven duplication if relevant.

Key drivers

Rank for cloud spend variance:

1. Compute hours (instance type, count, region) 2. Storage growth (class, lifecycle) 3. Egress (architecture-dependent) 4. Commit utilization and true-up 5. New services (AI/GPU, analytics) 6. FX (multi-currency billing) 7. Labor (if hybrid ops included)

Document correlation — drivers rarely move independently.

Sensitivity tornado

One-at-a-time % change on base annual spend:

driver: +10% API traffic → +$X (+Y%)
driver: −5% rightsizing → −$Z

Shows where leadership should focus governance vs accept variance.

Model hygiene

• Version models; date stamp assumptions
• Separate one-time from run-rate
• Link assumptions to named owner (product, infra, finance)
• Reconcile base case to finops-analyst trailing 12m actuals
• Avoid false precision — round to meaningful $ for exec audience

When model diverges >15% from actuals for 2 quarters, refresh assumptions before next commit decision.

TCO and build vs buy

Table of contents

1. TCO components 2. Cloud vs on-prem 3. Managed vs self-managed 4. NPV checklist

TCO components

Include all material costs over horizon (typically 3–5 years):

Do not compare invoice line to depreciation only.

Cloud vs on-prem

When cloud TCO wins (often):

• Variable or uncertain demand
• Short horizon or experimental workload
• Need global regions quickly
• Want to avoid hardware refresh cycle

When on-prem/colo may win:

• Stable 24/7 high utilization for years
• Very high scale with efficient operations team
• Strict data gravity with cheap local power
• Existing sunk hardware

Build crossover chart: cumulative cost vs time; note breakeven year.

Managed vs self-managed

Economic trade-off:

Model FTE hours × rate explicitly for self-managed options.

NPV checklist

1. Define discount rate (WACC or finance-provided) 2. Cash flows by year (CapEx, OpEx, migration) 3. Terminal value or explicit end-of-horizon exit cost 4. Sensitivity on utilization and growth ±20% 5. Document non-monetary factors separately (agility, compliance) — do not hide in fudge factors

Legal/regulatory constraints are inputs, not overridden by NPV alone.