💰 WHAT WE THINK

FinOps: What We Think

FinOps is not about cutting cloud costs. It's about aligning cloud consumption with the value it generates. The difference between those two sentences is the difference between a cost-reduction project that destroys operational capacity and an organizational practice that sustains growth.

Financial analysis center with multiple cloud data screens

Three illuminated data panels representing FinOps maturity levels

FinOps maturity: where most companies are (and where they should be)

The FinOps Foundation defines three maturity levels: Crawl (start measuring), Walk (actively optimize) and Run (continuously predict and optimize). In our experience, most are at advanced Crawl or early Walk: visibility of aggregated costs, some level of tagging and some monthly review process. But very few have costs attributed to business units with enough precision to make architecture decisions based on unit cost.

The most common trap is confusing activity with progress. A team can generate weekly reports, have monthly review meetings and a very detailed Grafana dashboard — and not have made a single relevant optimization decision in six months. Maturity is measured in actions with quantified impact, not in established processes.

Cost culture: why the problem isn't technical

The biggest obstacle to implementing FinOps is not the tool or the tagging taxonomy — it's the lack of real accountability. In most companies, cloud cost appears as a single line in the IT P&L without breakdown by product, team or feature. The engineers who make the technical decisions that determine spending don't see the economic impact of those decisions. Not out of negligence — because the system isn't designed for them to see it.

Building cost culture means every product team has visibility of their cloud spend in their usual operational dashboard, not in a separate report reviewed by someone in finance. It means that in sprint retrospectives, the unit cost impact of the cycle's changes is reviewed, just as velocity and defects are reviewed. The most effective change we've seen is creating the figure of the FinOps practitioner embedded in product teams, present in their rituals. That proximity is what makes the conversation about costs happen at the right moment.

Engineer reviewing cloud cost hologram and growth chart in server room

Real cloud ROI: dismantling the myth of automatic savings

The cloud doesn't reduce costs by default — it transforms them. Real ROI is built in three dimensions that are rarely calculated together: economic efficiency (cost per unit of capacity), business speed (time-to-market enabled by elasticity) and resilience (reduced incident cost). Companies that calculate only the first usually conclude that cloud is expensive. Those that calculate all three have a radically different perspective.

An exercise we do with clients: calculate the cost of engineering hours invested in managing on-premise infrastructure — patching, hardware failures, capacity planning — that in cloud are eliminated or radically reduced. For mid-sized companies, that cost frequently exceeds the price differential of resources. The honest TCO comparison includes the cost of human time, not just the vendor bill.

The most common mistakes: what we see in real projects

The first mistake is late tagging. Organizations usually start thinking about tagging strategy when they already have hundreds of resources without tags, or with inconsistent tags that make cost attribution impossible. Implemented from the start, as part of base IaC modules, it costs a fraction of that effort.

The second mistake is overprovisioning by default. In cloud, the natural tendency of engineers coming from on-premise is to provision with ample margin. AWS Compute Optimizer or Azure Advisor rightsizing recommendations are valid starting points, but require business context to apply with judgment.

The third mistake — the most costly — is not committing to Reserved Instances or Savings Plans when base consumption is stabilized. Organizations running predictable workloads 100% on-demand are paying a 30-60% premium over what they would pay with 1-year commitments.

Let's talk?

Do you know what each product really costs in the cloud?

We do a complete FinOps diagnosis: cost visibility, waste analysis, commitment strategy and cost culture roadmap for your organization.