When companies first engage with FinOps, they frame it as a cost-cutting exercise. The bill is too high. Someone needs to fix it.
This framing produces a particular kind of result: a short-term reduction in cloud spend, followed by a slow creep back to the previous level, followed by another cost-cutting initiative. The cycle repeats.
The underlying problem isn’t that the bill is too high. The problem is that nobody in the organisation owns it.
What Ownership Looks Like
A cloud bill you own is one where:
You know why every significant line item exists. Not at the account level — at the service and resource level. “We spend $47,000/month on EC2” is not ownership. “We spend $47,000/month on EC2 because the API servers run on c5.2xlarge and we’ve chosen on-demand for flexibility while we validate the traffic pattern” — that’s ownership.
Engineering teams see their costs. Developers making infrastructure decisions should know what those decisions cost. When a team chooses a managed database service over self-hosting, they should be able to see the cost difference and decide whether the operational simplicity is worth it. When cost is invisible, teams optimise for everything except cost.
Anomalies surface automatically, not accidentally. A 30% spike in data transfer costs shouldn’t be discovered when the bill arrives. It should be detected within 24 hours, attributed to a specific service or deployment, and investigated before it compounds.
Commitments are intentional. Reserved Instances, Committed Use Discounts, Savings Plans — these are financial commitments that require deliberate analysis. Buying them impulsively wastes money (wrong instance type, wrong term, wrong region). Avoiding them entirely also wastes money. The right answer requires understanding your workload patterns.
The Cultural Shift
The phrase “cloud costs” implies costs are a finance problem. They’re an engineering problem.
Every architectural decision a software engineer makes has a cloud cost implication. The database engine they choose. The caching strategy. Whether they paginate an API response or return all records at once. Whether they clean up temporary files or leave them in S3.
FinOps doesn’t work if it’s owned by a separate team that reviews bills and sends recommendations. It works when engineers treat cost as a first-class quality attribute of the systems they build — alongside performance, reliability, and security.
That cultural shift is hard to mandate. It’s easier to enable: give engineering teams cost dashboards, attribute costs to their services, make the data available without asking finance for a report.
The Vendor Relationship
SLA credits are a good lens for the vendor relationship dynamic.
When you file a credit claim, you’re holding a cloud provider accountable to a commitment they made in writing. Most cloud vendors process legitimate claims — they have to, it’s in the contract. But the process is designed to require effort on your part: you have to detect the breach, document it, file within the window, and follow up.
This is not accidental. The filing friction is a feature from the vendor’s perspective.
Owning your cloud spend means treating vendor commitments with the same rigour as your own. That includes understanding your SLAs, monitoring adherence, and filing claims when they’re breached. A company that never files SLA claims either has perfect cloud reliability (unlikely) or hasn’t built the process to capture what they’re owed.
What Changes When You Own It
You make better buy vs build decisions. When a team can see that their managed database costs $8,000/month and a self-hosted equivalent would cost $2,500/month in compute (plus engineering time), they can make an informed choice. When the cost is invisible, they usually default to managed because it’s easier — which is sometimes right and sometimes very expensive.
Architectural decisions get better. When an engineer knows that synchronous calls to a downstream service add $200/month in request costs at current traffic, they’ll think harder about caching, batching, and request optimisation. When it’s invisible, they won’t.
You negotiate better. A company that walks into enterprise discount negotiations knowing their exact spend by service, their commitment history, and their growth trajectory negotiates from a position of information. A company that brings a quarterly report from finance does not.
You retain the right to be surprised. When you own your costs, an unexpected bill is information — something changed that you didn’t expect. When you don’t own your costs, an unexpected bill is just a problem. The difference is whether you have the context to diagnose it.
The Goal Is Not a Lower Number
The goal of FinOps is not to minimise your cloud bill. It’s to maximise the value you get from every dollar you spend on cloud infrastructure.
Sometimes that means spending more — because you’ve identified a commitment that reduces unit costs, or because you’re expanding into a new region to improve latency for a customer segment.
Sometimes it means spending less — because you’ve found idle resources, or because a managed service you’re paying for isn’t delivering the value you expected.
The metric isn’t the bill. It’s the ratio of customer value delivered to infrastructure cost incurred — and your ability to explain, defend, and improve that ratio over time.
That’s ownership. That’s FinOps.
Fintropy is a multi-cloud FinOps platform in private beta. We help engineering teams own their cloud spend across AWS, Azure, and GCP. Learn more at nuvikatech.com