A deeper dive on how costs work

If you’ve found the service that makes up most of your Azure bill and now need to explain why it moves, this episode is for you. In the second episode of FinOps Thinking with Frank Contrepois: The Azure Sessions, FinOps expert Frank Contrepois takes the newcomer’s seat again and asks the questions anyone new to Azure pricing would ask. Host Michael Stephenson answers them live, in the Azure portal and in Turbo360.

Together they go one level below the invoice: what a meter is, why a virtual machine’s cost is never just the virtual machine, how storage, networking and App Service are charged, and when the meters are worth your attention. They close with the single lever that beats every discount for non-production workloads: turning things off.

What you’ll learn

• Think in rent plus meters – Every Azure resource has a fixed charge for being allocated and consumption meters for what you do with it.
• Cost the whole resource – Disks, networking, public IPs and logging all bill alongside a VM, and resource groups are how you see the total.
• Know the storage and PaaS levers – Redundancy, access tiers and lifecycle rules for storage; shared plans and the attribution problem for App Service.
• Spend attention where behaviour changes – Leave flat resources alone and dig into meters when a resource fluctuates or trends upwards.

Key Takeaways from the Episode

• A meter is a charge, not a resource – A VM that costs 50 pounds a month is really compute, licence and data transfer meters. Group by meter category and subcategory in Cost analysis to see them.
• The total cost of a VM is more than the VM – Disks, network interfaces, public IPs and monitoring agents bill separately. Logging into Log Analytics is a hidden cost of compute, and security and FinOps will want different levels of it.
• Work at the right level – FinOps, finance and procurement work at subscription, resource group and service level. Engineers work at the resource. Let an alert or a trend send you to a single resource.
• Storage savings sit in configuration – Capacity, transactions, egress and redundancy drive the bill. Access tiers and a lifecycle rule such as “not modified in 365 days, move to cool” are among the easiest savings to set up.
• Shared compute creates an attribution problem – App Service plans bill per node per hour, and Azure cannot tell you which app used what share. Decide whether you need cost per app before you consolidate plans.
• Turn things off before you optimise rates – A VM running eight hours a day on weekdays is allocated for 40 of 168 hours. Deallocate it properly, and remember that disks, public IPs and reservations keep billing while it is off.

In the episode, Michael also shows the Cost Driver Analysis agent in Turbo360 Cost Analyzer. From a single resource, it explains what each meter means, which levers matter most, and what would push next month’s bill up or down.

Host and guest

Enjoyed This Conversation?

Don’t miss other episodes of the FinOps on Azure Podcast, where we speak with practitioners, consultants, and cloud leaders about real-world cost management challenges, lessons learned, and what actually works in practice.

Subscribe for upcoming episodes and if this episode resonated with you, feel free to share it with your team.

Explore More FinOps Resources

Watch Webinars

Learn directly from FinOps practitioners and Azure experts in our Webinar Library.

Explore Case Studies

See how organizations and MSPs manage Azure costs at scale: Customer Stories.

Browse Resources

Access our full Resource Library – blogs, guides, videos, and best practices.