Every time the subject is cutting IT costs, companies' first reaction is to cut capacity. Shut down servers, reduce licenses, freeze hiring. The result is almost always the same: operations suffer, teams become overloaded, and in six months the cost comes back — often higher than before.

There is a different path. It has a name, a methodology, and proven results. It's called FinOps.

In recent years, working with clients such as BTG, B3, XP, and Inter, I witnessed firsthand how the FinOps discipline transformed these organizations' relationship with their technology spending. I'm not talking about marginal savings of 5% or 10%. I'm talking about reductions of 30%, 40%, in some cases 50% of total cloud costs — without shutting down anything that mattered, without sacrificing performance, without compromising growth.

This article is for decision-makers. If you are a CEO, CTO, CIO, or founder and want to understand how cloud cost optimization works in practice — not in the theory of vendor slide decks —, keep reading.

The real problem: you don't have a cost problem, you have a visibility problem

Most companies that come to me with an "IT cost problem" actually have a visibility problem. They know how much they pay on the monthly bill, but they don't know why they're paying that amount.

In a 2024 Flexera survey, 82% of organizations identified cloud cost management as their biggest challenge. But when you ask how many of them have a structured process for understanding where that money goes, the number drops to less than 30%.

In Brazil, the scenario is even more critical. With the rise of the dollar and the accelerated expansion of cloud workloads — especially following the rush toward generative AI in 2023 and 2024 — many companies simply lost track of what is running where, for what purpose, and at what cost.

The first principle of FinOps is precisely this: you cannot optimize what you cannot see. Before any cuts, before any decisions, you need granular visibility into your spending. This is technical and organizational work at the same time — and it's where most initiatives fail due to lack of structure.

What FinOps really is (and what it isn't)

FinOps is not your cloud team trying to justify the bill to the CFO. Nor is it the finance team trying to understand why AWS costs increased 40% last quarter without prior notice.

FinOps is the practice of bringing financial accountability to the variable cloud spending model, creating a culture where engineers, architects, product, and finance teams work together to make informed decisions about where to spend, where to save, and where to invest.

The FinOps Foundation, which maintains the most widely adopted framework in the market, defines three main phases:

  • Inform: create visibility and cost allocation by team, product, or cost center
  • Optimize: identify and implement cost reduction opportunities without operational impact
  • Operate: establish continuous governance and improvement processes

What sets FinOps apart from a simple "cost-cutting" initiative is the continuous and collaborative nature of the process. It is not a project with a beginning, middle, and end. It is a cultural shift that, when well implemented, pays for itself repeatedly over time.

The six biggest money wasters in cloud in Brazil

Based on more than two decades working in cloud computing and hundreds of reviewed architectures, the patterns of waste are surprisingly consistent. Regardless of the sector — financial, retail, healthtech — the culprits tend to be the same.

  • Oversized instances: servers running at 10% to 20% of actual utilization. It's the most common waste and the easiest to fix with the right data in hand.
  • Development and test environments running 24/7: no developer works 24 hours a day. Non-production environments left on outside business hours can account for 20% to 30% of the total bill.
  • Orphaned resources: storage volumes, elastic IPs, snapshots, and load balancers that aren't connected to anything useful but continue to be billed every month.
  • Poorly planned data transfer: data egress on AWS has a cost, and architectures that weren't designed with this in mind can generate significant and invisible charges.
  • Absence of Reserved Instances or Savings Plans: paying everything on the on-demand model when you have predictable workloads is leaving money on the table. Savings Plans can reduce costs by up to 72% for stable workloads.
  • Lack of tagging: without consistent tags, it's impossible to know which product, team, or client is generating which cost. Without this visibility, all optimization is guesswork.

At a financial sector client I recently worked with, just the first three items on this list represented R$ 1.2 million per year in waste. The remediation project lasted four weeks and cost a fraction of that.

AWS Cost Optimization in practice: where to start

If your main operation is on AWS — and for most medium and large Brazilian companies, it is — you have a robust set of native tools available for AWS cost optimization. The problem is not a lack of tools. It's a lack of process to use them.

The most efficient starting point I've ever found is AWS Cost Explorer combined with AWS Compute Optimizer. The first gives you the historical view and spending trends. The second uses machine learning to identify oversized resources and tells you exactly what the right size would be for each workload based on actual usage.

In parallel, AWS Trusted Advisor — available at the Business and Enterprise support tiers — continuously scans your environment and highlights recommendations for cost, security, and performance. It's like having a technical consultant running 24 hours a day on your account.

For companies with multiple AWS accounts — which is the reality of any well-structured mid-sized organization — AWS Organizations with consolidated billing allows you to view aggregate spending and apply volume discounts. Many companies miss this benefit simply by not having organized their accounts properly.

The rule I use with all my clients: before buying more capacity, prove that you're making good use of what you already have. In 90% of cases, the capacity already exists — it's just poorly allocated.

One concrete action that can be implemented in less than 30 days: set up budgets with automatic alerts in AWS Budgets for each team or product. When people start receiving notifications about the costs they generate, behavior changes naturally. This is FinOps in its simplest form — and it works.

The organizational dimension: why FinOps fails without executive sponsorship

Here is an uncomfortable truth: most FinOps initiatives fail not due to a lack of technology, but due to a lack of organizational alignment.

The engineering team wants the freedom to provision resources quickly. Finance wants predictability and control. Product wants delivery speed. These objectives may seem conflicting, but they aren't — when there is a common framework and an executive sponsor who ensures everyone is playing on the same team.

In the financially efficient IT management implementations I've seen truly succeed, there is always a person (or function) acting as a bridge between the technical world and the financial world. Some companies call this role a FinOps Practitioner. Others incorporate this responsibility into the CTO or a senior architect. The name doesn't matter. What matters is that someone exists with the authority and context to translate technical decisions into financial impact — and vice versa.

Another critical element is building a "you build it, you pay for it" culture. When engineering teams are held accountable for the cost of what they build — not in a punitive way, but as a quality metric — architecture decisions change. An engineer who knows their feature will cost R$ 50,000 per month in infrastructure makes different choices than an engineer for whom that cost is invisible.

Real results: what to expect from a well-executed FinOps initiative

I'll be straightforward about numbers, because I believe realistic expectations are more useful than inflated promises.

In a first well-structured FinOps initiative, companies typically find between 20% and 35% of immediate reduction opportunity — that obvious waste that no one was seeing. This is the so-called "low-hanging fruit."

In a second cycle, with a continuous process established and engaged teams, it's common to find an additional 10% to 20% optimization through better architectural choices, use of spot instances for interruption-tolerant workloads, and capacity commitment with Reserved Instances.

To make it concrete: for a mid-sized company with a monthly cloud bill of R$ 500,000, we're talking about a potential savings of R$ 100,000 to R$ 175,000 per month. Over 12 months, that's more than R$ 1 million returned to the business — without cutting any capacity that matters, without compromising any delivery.

One point few people talk about: FinOps is not just about saving, it's about spending better. In some cases, the right outcome of a FinOps initiative is to increase investment in certain areas — because now you have evidence that the spending generates returns. The discipline creates confidence for the CFO to approve technology investments that would previously have been denied due to a lack of visibility into ROI.

Where to start tomorrow

If you've made it this far and are wondering where to begin, I'll give you a concrete 90-day roadmap.

In the first 30 days, focus on visibility. Activate Cost Explorer, implement a mandatory tagging policy for all new resources, configure Compute Optimizer, and generate the first report of oversized resources. This work will give you a true picture of the problem.

Between 30 and 60 days, implement the low-risk, high-impact optimizations: shut down non-production environments outside business hours, eliminate orphaned resources, and right-size the most obviously oversized instances. Document the savings generated — you'll need those numbers for the next step.

Between 60 and 90 days, present the results to executive leadership and propose the establishment of a continuous process. With the numbers in hand, the conversation about formalizing the FinOps role and creating regular cost review rituals becomes much easier.

Sustainable technology cost reduction doesn't come from one-off cuts. It comes from building an organization that continuously understands, questions, and optimizes how it spends on technology. That's the difference between companies that stall their growth trying to save money and companies that scale with efficiency.

If your company still doesn't have this structured process — or if it does, but isn't generating the expected results — it's worth having a conversation. Over more than 20 years working with the largest organizations in Brazil's financial and technology sectors, I've developed a way to diagnose and solve these problems quickly and without operational impact.

Get in touch at abraao.tech. A 30-minute conversation could be the beginning of significant savings — and a much smarter IT operation.