Your first Azure cost optimization sprint should run two weeks and follow a strict order of operations. Take the zero risk quick wins first: deallocate idle resources, delete unattached managed disks, and schedule non production environments to switch off out of hours. Then rightsize oversized VMs against real utilization. Only then layer Azure Reservations, the Azure Savings Plan, and Azure Hybrid Benefit on the steady state. Doing commitments before rightsizing locks in the wrong size, so sequence matters more than effort.
This is an Azure cost fundamentals exercise. Here is week by week what to do, the numbers to capture, and the traps that waste a first sprint.
Start from Azure cost exports, not the portal summary
Begin with Azure cost exports in the FOCUS format, the standardised billing schema, so the sprint reconciles against granular data rather than a dashboard average. Azure Cost Management is fine for a first look, but the export is what lets you attribute spend to subscriptions, resource groups, and tags. If tagging is thin, the first half day of the sprint is fixing it, because every later decision depends on knowing who owns what. The fuller picture lives in the Azure cost optimization guide.
Week one: the quick wins
Week one is discovery plus the savings that carry no performance risk and need no engineering debate.
- Deallocate idle resources. VMs running at near zero utilization, often forgotten test environments. Stopped and deallocated VMs stop billing for compute.
- Delete unattached disks. Managed disks left behind when VMs were deleted bill every month for nothing.
- Schedule non production off. Dev, test, and staging rarely need to run nights and weekends. Auto shutdown schedules cut those environments by well over half of their hours.
- Catch the quiet eaters. Over retained Log Analytics data and ungoverned bandwidth charges, which rarely show up in a headline view.
Week two: rightsizing, then commitments
Week two is the working half, run with the engineering teams in the room.
Rightsizing comes first. Azure Advisor recommends candidates, but it recommends, it does not decide, so confirm each move against real workload behaviour before resizing. Get the fleet to its true size before you commit to anything, because a reservation bought around an oversized VM bakes the waste in for a year or three.
Then commitments. With the fleet rightsized, size Azure Reservations and the Azure Savings Plan to the steady state floor your forecast can defend, and check Azure Hybrid Benefit eligibility first, because applying existing Windows Server and SQL Server licences changes which commitment is worth buying. Coverage follows a defensible forecast, not the largest available discount. If you run an Enterprise Agreement, factor MACC drawdown into the timing so commitments help you burn the balance rather than stranding it.
A European SaaS company ran its first Azure sprint and assumed reservations were the headline lever. The cost export told a different story: idle VMs and unattached disks were a steady monthly leak, non production ran 24 by 7, and a third of the fleet was a size too large. Quick wins and rightsizing landed in week one and two, and only then did right sized reservations plus Hybrid Benefit on the SQL estate lock in the rest. Figures are verified against billing data and anonymised.
What turns a sprint into a flat bill?
A sprint that ends with no owner reverts. Before you close it, assign the recurring checks: an idle resource and unattached disk report in the monthly review, an auto shutdown policy as the default for new non production, and a quarterly commitment review against the forecast. This is the discipline that holds the 31 percent median reduction we see in the first 90 days rather than letting it leak back. The standing rhythm is covered in the Azure spend review cadence, and the recurring waste categories in the biggest Azure waste categories.
Frequently asked questions
Run your first Azure sprint with us
We run the first Azure cost sprint with your teams, bank the quick wins in week one, and size commitments to a forecast your engineers will stand behind. Our guarantee: we reduce your cloud spend or we reimburse our service fee.
Put a defensible number on your cloud spend.
No provider in the room, no published price list. Tell us your footprint and we will scope the savings against your billing data — we reduce your cloud spend or we reimburse our service fee.
The Cloud Spend Navigator: what changed in cloud pricing, commitments, and FinOps — no vendor spin.