Reduce Azure waste in the right order: establish ownership, remove idle capacity, rightsize workloads, then assess reservations or savings plans against stable demand.
Topic Azure
Published 15 June 2026
Updated 20 July 2026
Summary
Remove unused resources, schedule non-production systems and rightsize active workloads before you buy reservations or savings plans. Measure the result against a clear cost baseline.
Context
Azure charges reflect architecture and operating decisions made across subscriptions, resource groups and teams. Costs get hard to control when resources lack owners, budgets don’t reflect expected usage, or technical teams treat the monthly invoice as a finance problem.
Cost optimisation is ongoing operational work. It should protect reliability, security and delivery commitments while removing spend that doesn’t support a business outcome.
Why it matters
Poor cost visibility delays decisions. An unused disk, an oversized database or a test environment that never stops might look minor alone, but the gaps add up across an estate.
The reverse problem matters too. A rushed cost cut that ignores performance, recovery or licensing creates operational risk. Every recommendation needs an owner, evidence and a rollback path where it changes service capacity.
Risks and considerations
Discounts can preserve waste
Reservations and savings plans reduce eligible rates in exchange for a one-year or three-year commitment. They work best for stable, understood demand. Rightsizing or shutdown changes that demand, so calculate commitments after those changes settle.
Low utilisation needs context
A quiet virtual machine might provide standby capacity, support a monthly process or meet a recovery requirement. Review utilisation over a period that covers the workload cycle, and speak with the owner before changing it.
Storage tiers have access costs
Moving data to cool, cold or archive tiers changes retrieval cost and availability. Confirm access patterns, retention obligations and recovery expectations before changing lifecycle rules.
Allocation is not optimisation
Tags and cost centres explain who pays. They don’t reduce spend by themselves. Use allocation data to create ownership and prompt decisions.
Recommended approach
Establish the baseline. Review actual and amortised cost, major service categories, recent changes and forecast demand.
Assign ownership. Record an owner, environment, service and cost centre for each material resource group or workload.
Remove obvious waste. Investigate unattached disks, old snapshots, unused public IP addresses, idle services and superseded test resources.
Schedule non-production workloads. Deallocate eligible virtual machines and other resources outside required hours. Confirm shutdown automation results in the billable state you expect.
Rightsize active services. Use Azure Advisor, platform metrics and workload knowledge. Change one controlled group at a time and monitor performance.
Review architecture and licensing. Check Azure Hybrid Benefit, service tiers, scaling rules and whether managed services reduce total operating cost.
Assess commitments. Compare reservations and savings plans only after demand is stable. Review scope, utilisation, term, exchange limits and ownership.
Measure realised savings. Record the recommendation, approved action, expected saving and observed cost after implementation.
Microsoft recommends applying Azure Advisor actions in this order: rightsizing or shutdown, then reservations, then savings plans. Each action changes the demand used to calculate the next recommendation.
Practical next steps
Run a focused review of your largest cost drivers:
identify the workload owner and business purpose
compare the last three complete billing periods
review Azure Advisor recommendations and utilisation data
check whether existing reservations or savings plans are fully used
agree one low-risk action and measure the result
Further reading
Calculate cost savings with Azure Advisorㅤ❯
Cost optimisation recommendations in the Azure Well-Architected Frameworkㅤ❯
Related case studies and services
ONRSR Azure governance case studyㅤ❯
Apply this guidance
Discuss how this advice fits your current Microsoft environment, risks, and priorities.