Azure Cost Optimisation Playbook: How to Cut Your Azure Bill by 35%

Azure Cost Optimisation: 35% Reduction Playbook

Most mid-market companies on Azure are overpaying by 25-40%. Not because Azure is expensive, but because the default configuration — pay-as-you-go pricing, no reserved instances, no auto-shutdown, no storage tiering — is optimised for flexibility, not cost.

This playbook covers the specific actions that consistently deliver the largest savings. Implement all of them and a 35% reduction is achievable within 90 days.

1. Reserved Instances (Typical saving: 40-72% on VM costs)

Azure Reserved VM Instances allow you to commit to 1 or 3 years of VM usage in exchange for a significant discount over pay-as-you-go rates. Typical savings: 1-year reservation: 40% off pay-as-you-go. 3-year reservation: 60-72% off pay-as-you-go.

Action: Identify all VMs running 24/7 with predictable workloads (production app servers, databases, ERPNext instances). Purchase 1-year reserved instances for these. For VMs with uncertain future sizing, use 1-year flexible scope reservations which allow instance size flexibility within a VM series.

Real example: A client with £8,200/month in VM costs (pay-as-you-go) reduced to £4,700/month after purchasing 1-year reserved instances for their production workloads. Saving: £42,000/year.

2. Azure Hybrid Benefit (Typical saving: 40-85% on Windows/SQL costs)

If you have existing Windows Server or SQL Server licences with active Software Assurance (SA), Azure Hybrid Benefit (AHB) allows you to use those licences on Azure VMs, paying only for the compute cost.

Windows Server VMs: AHB removes the Windows licence cost from the VM price (typically 40-50% of the VM cost for Windows VMs). SQL Server on Azure SQL MI or Azure VM: AHB removes the SQL licence cost — savings of 55-85% depending on SQL edition vs VM size.

Action: Check your Microsoft EA or MCA agreement for Software Assurance coverage. Enable AHB on all eligible Windows and SQL VMs in the Azure portal or via Azure CLI.

3. VM Rightsizing (Typical saving: 15-30% of VM costs)

VMs are routinely overprovisioned. The standard pattern: a developer requests a D4s v3 (4 vCPU, 16GB RAM) for a workload that uses an average of 12% CPU and 4GB RAM. A D2s v3 (2 vCPU, 8GB RAM) would be sufficient and costs half as much.

Action: In Azure Advisor, review the Cost recommendations tab. Azure Advisor automatically identifies VMs with low average CPU utilisation and recommends downsizing. Apply recommendations to non-production environments first, validate performance, then apply to production.

4. Dev/Test Auto-Shutdown (Typical saving: 60-70% of dev VM costs)

Development and test VMs running 24/7 is one of the most common Azure waste patterns. A VM used 8 hours/day, 5 days/week is running 168 hours/week but only used for 40 hours — 76% waste.

Action: Enable Auto-shutdown on all dev/test VMs (set to shut down at 7pm local time). Use Azure DevTest Labs or Azure Automation runbooks to auto-start VMs at the beginning of the working day. Implement Start/Stop VMs during off-hours solution for environments that need scheduled availability.

5. Storage Tiering (Typical saving: 40-80% on cold data storage)

Azure Blob Storage pricing by tier: Hot: £0.018/GB/month (frequently accessed). Cool: £0.009/GB/month (infrequently accessed, 30-day minimum). Cold: £0.004/GB/month (rarely accessed, 90-day minimum). Archive: £0.001/GB/month (long-term retention, 180-day minimum, rehydration cost applies).

Action: Implement Azure Blob Lifecycle Management policies to automatically tier objects based on last-accessed date. Typical rule: move to Cool after 30 days, Cold after 90 days, Archive after 365 days. Review managed disk snapshots — these are frequently left on Premium SSD when Standard HDD is sufficient for backup snapshots.

6. Identify and Delete Orphaned Resources

Every Azure environment accumulates orphaned resources: unattached managed disks (VMs deleted but disks left), unused public IP addresses, empty load balancers, unused App Service Plans, stale snapshots. These cost money every month for zero value.

Action: Run Azure Resource Graph queries to identify unattached disks, unused public IPs, and empty App Service Plans. Delete or deallocate confirmed orphaned resources. Implement a tagging policy (owner, environment, cost-centre) to prevent future orphans.

Implementation Sequence for Maximum Impact

Week 1: Enable Azure Hybrid Benefit on all eligible resources (zero risk, immediate saving). Week 2: Implement auto-shutdown on all dev/test VMs. Week 3: Delete orphaned resources. Week 4: Implement storage tiering lifecycle policies. Weeks 5-8: Complete rightsizing assessment and apply to non-production. Month 3: Purchase reserved instances based on validated production sizing.

Techseria runs Azure cost optimisation engagements that typically deliver 25-40% cost reductions within 90 days. Book a Strategy Session — we'll audit your current Azure spend and identify specific saving opportunities before you commit to anything.