The FinOps Framework: Inform, Optimize, Operate

The FinOps Framework is the vendor-neutral reference model maintained by the FinOps Foundation. It has three primary elements:

• Principles — the values that guide the practice
• Domains and Capabilities — the work that gets done
• Maturity — Crawl, Walk, Run for each capability

The framework is structured around three iterative phases teams move through continuously: Inform, Optimize, Operate.

The Six Principles

1. Teams need to collaborate. FinOps is a team sport between engineering, finance, business, and FinOps practitioners.
2. Everyone takes ownership for their cloud usage. Engineers see and act on their own costs.
3. A centralised team drives FinOps. A small FinOps team enables the rest of the organisation but does not own optimisation alone.
4. FinOps reports should be accessible and timely. Hours-to-days latency at most; visibility in the tools engineers already use.
5. Decisions are driven by business value of cloud. Cost is one input alongside performance, reliability, time-to-market.
6. Take advantage of the variable cost model of cloud. Use elasticity, commitments, and architectural flexibility — don't try to make cloud look like on-prem.

You will see these reflected in the rest of the framework.

The Three Phases

Inform — visibility, allocation, benchmarking

You cannot optimise what you cannot see. The Inform phase is about getting the right data to the right people:

• Visibility of spend — by service, account, environment, team, product
• Allocation of shared costs — what does Team A owe, what does Team B owe
• Tagging and labelling — the prerequisite for allocation
• Benchmarking — internal trends and external industry benchmarks
• Forecasting — what will next month cost
• Anomaly detection — spikes and drift
• Budgeting — targets and alerts
• Showback / chargeback — making the spend visible (showback) or actually debiting it from team budgets (chargeback)

This phase has the lowest engineering effort and the highest leverage. Most savings opportunities are discovered here even though the savings are realised in Optimize.

Optimize — taking action

Acting on the insights from Inform:

• Rightsizing — match resource size to actual demand
• Eliminate waste — idle resources, orphaned storage, unused load balancers
• Commitment-based discounts — RIs, Savings Plans, CUDs
• Storage tiering — Standard → Infrequent Access → Glacier
• Architectural optimisations — Spot instances, serverless where cheaper, consolidating accounts
• Workload optimisation — code-level changes to reduce CPU, memory, IO
• Egress and network optimisation — CDNs, private endpoints, traffic locality

This is the phase that delivers numbers to finance. Most courses, vendors, and conference talks live here. But Optimize without Inform is whack-a-mole, and without Operate the gains erode.

Operate — making it stick

The cultural and governance phase. Without it, every optimisation campaign decays:

• FinOps team and structure — who owns this, where do they sit
• Cadence — weekly stand-ups, monthly reviews, quarterly business reviews
• Policies — tagging standards, budget thresholds, anomaly thresholds
• Automation — auto-shutdown, auto-rightsizing, auto-tagging enforcement
• Education — engineers learn FinOps; finance learns cloud
• Tooling integration — cost in CI, cost in monitoring, cost in product analytics
• Continuous improvement — retrospectives, KPIs, maturity progression

The Cycle Is Continuous

Phases are not sequential project gates. Mature teams iterate all three weekly:

• Monday — Inform: review last week's spend report and anomalies.
• Tuesday-Thursday — Optimize: act on flagged items.
• Friday — Operate: update policies, retrospectives, maintain commitments.
• Repeat.

The Capabilities

FinOps Foundation maps 22 capabilities (in the framework v2 of 2024) to the three phases and four "domains" (Understand Usage and Cost, Quantify Business Value, Optimize Usage and Cost, Manage the FinOps Practice). Some of the most common:

The full list is on finops.org with crawl/walk/run rubrics per capability.

Maturity Levels

Mature organisations sit between Walk and Run on most capabilities. "Run" everywhere is aspirational — pick the capabilities that matter for your business model.

The Order to Build

For a brand-new FinOps practice the typical sequence:

1. Visibility. Connect billing data to a tool that engineers can access — even AWS Cost Explorer / Azure Cost Management / GCP Billing reports + a shared dashboard is enough to start.
2. Tagging standard. Define mandatory tags (cost-centre, environment, team, product). Enforce via Service Control Policies, Azure Policy, or organisation policy.
3. Allocation report. Monthly showback by team. Even imperfect is better than nothing.
4. Anomaly alerting. AWS / Azure / GCP native anomaly detection, free, useful.
5. Quick-win optimisations. Idle resource cleanup, commitments for steady workloads.
6. Cadence. Weekly FinOps stand-up. Monthly review with finance and engineering leaders.
7. Iterate. Each cycle adds maturity in one or two capabilities.

This is what the remaining lessons walk through in depth — starting with visibility and allocation, the prerequisites for everything else.