FinOps Principles Applied to Azure

The three FinOps principles

The FinOps Foundation defines cloud financial management through three iterative phases. Teams work through all three continuously — not sequentially.

Inform: building cost visibility in Azure

The Inform principle is about creating a shared, accurate view of cloud spend before attempting to optimise it. You cannot optimise what you cannot see, and you cannot allocate what you cannot attribute.

Tagging for cost allocation. The foundation of Azure cost visibility is resource tagging. Without tags, Cost Analysis can show you total spend by resource type or subscription, but cannot attribute that spend to the team, project, or environment that generated it. A minimal effective tagging strategy includes:

• environment — values: prod, staging, dev, test
• team — the engineering or business team that owns the resource
• project — the workload or initiative the resource supports
• cost-centre — the internal accounting code for chargeback (optional but valuable at scale)

Cost Analysis dashboards. Azure Cost Analysis (in the Azure portal under Cost Management + Billing) provides built-in views for spend over time, grouped by any tag. Once tagging is consistent, you can build a shared dashboard that shows each team’s spend trend, their largest cost drivers, and their spend vs budget position.

Exporting billing data. For deeper analysis, export billing data to Azure Blob Storage on a daily or monthly schedule. The exported CSV or Parquet files contain line-item detail — every resource, its cost, its tags, and its usage quantities. Teams with a data warehouse (Synapse, Power BI) can build custom reports from this data that the Cost Analysis portal cannot produce.

# Create a billing export to a storage account
az costmanagement export create \
  --name monthly-billing-export \
  --scope /subscriptions/{subscription-id} \
  --type ActualCost \
  --timeframe MonthToDate \
  --recurrence Monthly \
  --recurrence-period from="2026-04-01" to="2027-04-01" \
  --storage-account-id /subscriptions/{sub-id}/resourceGroups/{rg}/providers/Microsoft.Storage/storageAccounts/{account} \
  --storage-container billing-exports \
  --storage-directory azure-costs

Optimise: reducing spend in Azure

With visibility established, the Optimise principle focuses on acting on what the data reveals. Azure offers several well-defined levers for cost reduction.

Azure Advisor recommendations. Advisor is the starting point for optimisation. Under the Cost category, it surfaces rightsizing opportunities, idle VMs, unattached disks, and reserved instance purchase recommendations. Each recommendation includes an estimated annual saving. Address high-impact items first.

Reserved Instances. For predictable, always-on workloads — production databases, application servers, Kubernetes nodes — purchasing 1-year or 3-year Reserved Instances provides significant savings over pay-as-you-go pricing. A 1-year reservation on a Standard_D4s_v5 in East US saves approximately 36% compared to pay-as-you-go; a 3-year reservation saves approximately 56%. Savings Plans provide similar discounts with more SKU flexibility but slightly lower maximum savings.

Spot VMs for batch and fault-tolerant workloads. Spot VMs use excess Azure capacity at up to 90% discount vs pay-as-you-go but can be evicted with 30 seconds’ notice. They are suitable for batch processing, CI/CD pipelines, rendering, and ML training jobs where interruption can be handled gracefully.

Idle resource cleanup. Regularly removing unattached disks, orphaned public IPs, empty App Service Plans, and unused load balancers eliminates pure waste with zero performance trade-off.

Operate: sustaining cost discipline at scale

The Operate principle is about building systems and habits that keep cost management working without requiring constant manual effort.

Budget alerts. Azure budgets let you set a spending threshold for a subscription, resource group, or tag scope, and trigger email or webhook alerts when spend reaches a defined percentage (e.g., 80%, 100%, 110% of budget). Budgets do not automatically stop resources — they are alerts only — but they ensure the right people know when spend is trending over target in time to act.

Azure Policy for cost governance. Policy can enforce tagging requirements at resource creation time, prevent deployment of expensive resource types in dev subscriptions, and require resources to be within approved regions. These guardrails reduce the number of cost surprises that need to be cleaned up reactively.

Chargeback and showback workflows. Chargeback transfers financial accountability to the team or department that generated the spend. Showback reports cost data to teams without transferring financial liability. Both require accurate tagging. In practice, most organisations start with showback — making teams aware of their spend — before moving to chargeback, which requires accounting system integration.

A practical chargeback workflow in Azure:

1. Export billing data daily to a storage account.
2. Use Power BI or a Synapse query to aggregate costs by the team and cost-centre tags.
3. Distribute a monthly cost report to each team lead showing their total spend, top 5 cost drivers, and month-over-month trend.
4. Raise internal invoices or journal entries to the relevant cost centre for chargeback.

A practical monthly FinOps workflow

FinOps requires a regular cadence. Without a scheduled review, recommendations accumulate, budgets drift, and cost visibility degrades. The following is a realistic monthly workflow for a team managing Azure spend.

FinOps maturity levels in Azure

The FinOps Foundation defines three maturity levels: Crawl, Walk, and Run. Most teams start at Crawl and take 12–18 months to reach Run.

• Crawl: Basic tagging on major resources, budget alerts on top-level subscriptions, manual monthly cost review, ad-hoc Advisor action.
• Walk: Consistent tagging enforced by Policy, automated billing exports, showback reports to team leads, scheduled Advisor reviews, first Reserved Instance purchases for stable workloads.
• Run: Full chargeback to business units, automated idle resource cleanup, Savings Plans or 3-year RIs for all predictable workloads, cost embedded into engineering team KPIs, anomaly detection alerts.