Commitment Portfolio Flexibility Index (CPFI)

GMI-IT-014IT OperationsIntegration-Required

Definition

Commitment Portfolio Flexibility Index (CPFI) is the proposed metric for this: the proportion of committed cloud spend that could be reallocated to a different instance family, region, or service within 90 days without financial penalty.

The Measurement Gap

Your FinOps platform is winning. RI utilization: 92%. Savings plan coverage: 85%. The dashboard is green. The executive summary says you've cut cloud waste by $2.1 million this year through AI-driven purchasing optimization.

Then your engineering team announces a container migration. The monolithic application that's been running on 40 m4.xlarge reserved instances for three years is moving to Kubernetes over the next 18 months. Those reservations have 2 years left on their terms. They're specific to that instance family. They can't follow the containers.

The FinOps AI did exactly what it was designed to do: it saw stable, high-utilization workloads and bought multi-year, instance-specific reservations to maximize discount depth. It optimized each purchasing decision locally. It never had visibility into the organization's architectural roadmap. And now you have $840,000 in committed spend attached to an infrastructure pattern that's being retired.

This is the failure mode that portfolio inflexibility creates: the AI purchases aggressively because aggressiveness looks good on the dashboard, and the portfolio becomes increasingly brittle without anyone measuring it. The architectural team isn't tracking commitment flexibility. The FinOps team isn't tracking it either: they're tracking utilization, coverage, and savings, all of which can look excellent while the portfolio is becoming less able to absorb change.

The legacy metrics don't see this because they're designed to measure what you bought and whether you're using it, not whether what you bought can adapt to what comes next. RI utilization at 92% tells you the commitments are being consumed. It says nothing about whether those commitments would survive a major architectural shift. Savings plan coverage at 85% tells you that most compute spend is discounted. It doesn't tell you what percentage of that discounted spend is locked into rigid instruments that can't be exchanged or converted without penalty.

The organizations that feel this most sharply are the ones making smart modernization moves: container migrations, serverless transitions, Graviton adoption, multi-cloud redistributions. Every one of those transitions leaves a trail of legacy commitments that the AI's optimization logic created. And because each individual commitment made sense when it was purchased, there's no obvious villain. The AI performed well. The FinOps team did their job. The problem is structural: nobody measured whether the cumulative purchasing strategy preserved the portfolio's ability to change.

The measurement gap for CPFI is this: there is no metric that tells you whether your current portfolio of cloud financial commitments can absorb a major architectural change without incurring significant penalty costs. Without that metric, the AI purchasing tool optimizes for discount depth, and gradually trades adaptability for savings, one well-reasoned commitment at a time.

Formula

Commitment Portfolio Flexibility Index (CPFI) is the proposed metric for this: the proportion of committed cloud spend that could be reallocated to a different instance family, region, or service within 90 days without financial penalty. CPFI is a portfolio-level measure, not a per-commitment measure. Individual commitments may be fully rigid or fully flexible. CPFI captures whether the portfolio as a whole has preserved enough optionality to absorb realistic architectural change.

Formula (plain language:

Take all active cloud financial commitments) reserved instances, savings plans, committed use discounts, enterprise discount programs. Classify each as flexible (exchangeable, convertible, or instrument-agnostic like a compute savings plan) or rigid (locked to a specific instance family, region, or service that can't be exchanged without penalty). Sum the dollar value of flexible commitments and divide by total committed spend. The result is the percentage of your committed portfolio that can be redirected without financial penalty.

Formula: notation:

CPFI = Σ C_flexible / Σ C_total

Where:

  • C_flexible = annual committed spend classified as convertible, exchangeable, or covered by instrument-flexible savings plans
  • C_total = total annual committed cloud spend across all instruments and providersWorked Example An organization has $3.2M in annual committed spend. Breaking it down: $1.4M is in AWS Compute Savings Plans (flexible. Applies across any EC2 family, any size, any region). $800K is in convertible RIs (flexible: can be exchanged for equal or greater value). $600K is in standard RIs on specific instance families (rigid). $400K is in Azure Reserved VM Instances for specific VM types (rigid). CPFI = ($1.4M + $800K) / $3.2M = $2.2M / $3.2M = 68.75%. That's a reasonably flexible portfolio. If the $1.4M in savings plans were standard RIs instead, CPFI would drop to 25%, a portfolio that can't absorb much change without pain.

Sign convention:

Higher is better. 100% would mean every commitment is fully exchangeable or convertible: theoretically possible, practically sub-optimal because flexible instruments carry smaller discounts. 0% means every dollar is locked to a specific instance type and region: achievable only through uniformly aggressive standard-RI purchasing.

Edge Cases

Multi-cloud portfolios should be calculated per provider and then aggregated: flexibility across cloud providers requires workload portability, not just commitment portability, and they can't be directly compared. Compute savings plans that cover some services but not database should have only the compute-eligible spend counted as flexible. Convertible RIs on AWS have exchange restrictions (equal or greater value requirement): count them as flexible but track the constraint. A portfolio with zero commitments should return CPFI as undefined, not 100%.

Cross-references:

Upstream: Commitment Recommendation Accuracy (CRA, GMI-IT-013). Low CRA on rigid instruments creates more stranded commitments and directly reduces CPFI over time. When the AI purchases the wrong rigid commitments, those purchases lock down flexibility with nothing to show for it. Downstream: Stranded Commitment Waste Rate (SCWR, GMI-IT-015). Low CPFI combined with architectural change creates the conditions for stranded commitments. A flexible portfolio can be redirected; a rigid portfolio gets stranded. Downstream: Arbitrage Concentration Risk (ACR, GMI-IT-017). Concentrated workloads with inflexible commitments are doubly constrained: they can't move to different providers because commitments don't transfer, and they can't change instance families because the RIs are instance-specific.

Proposed formula, pending validation pilots.

At a Glance

Category
AI-Era IT Operations
← Code Comprehension Debt Index (CCDI)Commitment Recommendation Accuracy (CRA) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment