Cloud and SaaS costs with FinOps
Mangrove's cloud bill is about $540,000 a month, and almost nobody can say who spends it. This lesson shows you how to read the bill, give every dollar an owner with tags and split rules, cut cost in a safe order, buy commitments without locking in waste, and track one number that tells growth apart from waste: cloud cost per order.
By the end you can
- Explain why cloud spend surprises finance, and name the FinOps phases and principles
- Read a cloud bill by service, usage type and pricing model, and spot data transfer costs
- Allocate cost to teams with tags, showback and a fair rule for shared and untagged cost
- Rank savings levers by safety and work out the money for scheduling and commitments
- Track cloud cost per order and run a simple SaaS seat and renewal review
The everyday picture
Think of a big shared house that pays one electricity bill. Nobody sees the meter while they use power, so a heater stays on all day, a fridge in the garage runs empty, and the bill climbs. When it jumps, everyone says it was someone else.
A house that gets this under control puts a small meter on each room, agrees how to split the hallway lights, switches off what nobody uses, and puts a timer on the garage heater. Then it signs a cheaper fixed-price deal for the power it always uses, and pays the normal rate for the extra on cold nights. That is FinOps: meters, fair splits, switching off, timers and the right tariff.
Why cloud spend surprises everyone
In a data centre, spending was slow and visible: a purchase order for servers, an approval, then an asset on the balance sheet for years (capex, from Budgets, capex and opex). Cloud changes this in three ways.
- It is variable. You pay for what runs, by the hour or gigabyte. A festival week costs more than a quiet one, and nobody signed anything.
- It is decentralised. Hundreds of engineers can start resources, so spending decisions sit in every team.
- A config change is a purchase. Moving an instance from large to extra large in a pull request can double that service's cost.
FinOps brings ownership to this world. The FinOps Foundation, part of the Linux Foundation, publishes the framework most companies use. Its three phases repeat as a loop:
- Inform: see the spend. Who uses what, how much, and what it delivers.
- Optimize: find and agree on ways to improve, both in how resources are used and in the rates paid for them.
- Operate: put the changes into daily work, measure them, and go back to Inform.
It also has six principles. In the Foundation's current wording: teams need to collaborate; business value drives technology decisions; everyone takes ownership for their technology usage; FinOps data should be accessible, timely and accurate; FinOps should be enabled centrally; and take advantage of the variable cost model of the cloud. The last one is easy to miss: a bill that goes up because orders went up can be good news. The framework also applies these practices beyond public cloud, to SaaS, licences and data centres, which it calls scopes.
Quick check
Which phase is a team in when it first builds a dashboard showing each team's monthly cloud cost?
Inform is about visibility and allocation, and a cost-per-team view is its core output. Savings come later in Optimize, once people can see their own numbers. Operate is where agreed changes become routine, such as a weekly review of that dashboard.