Organizations are still wasting 27% of their cloud spend, and according to the Flexera 2025 State of the Cloud Report their cloud budgets are overrunning by an average of 17%. That combination — money evaporating into underutilized resources plus budgets that blow out quarter after quarter — is exactly the problem FinOps was built to solve. FinOps is the discipline that turns that loss of control into measurable savings, aligning engineering, finance, and business around a single question: what value are we getting for every dollar we spend in the cloud?
If your organization has already migrated workloads to AWS, Azure, or Google Cloud and the monthly bill is growing faster than your ability to explain it, this guide is for you. We'll cover what FinOps really is (beyond the buzzword), how its framework is structured, where the real saving levers are across the three major cloud providers, and how to build a team and a culture that sustains those savings over time — not just during a one-off cost-cutting campaign.
What FinOps is and why it matters to your business
The FinOps Foundation defines FinOps as an operational framework and cultural practice that maximizes the business value of technology, enables data-driven decisions, and creates financial accountability through collaboration between engineering, finance, and business. That definition contains three ideas worth unpacking.
First, it is both operational and cultural. Buying a cost-visibility tool is not enough; FinOps changes who makes spending decisions and with what information. Second, the goal is not to spend less — it is to maximize value. Sometimes the right decision is to spend more on a revenue-generating workload, as long as it is a deliberate, traceable decision. Third, financial accountability is distributed: the engineer who deploys an instance should see and own the cost of that decision in near-real time, rather than discovering it on next month's invoice when it is already too late.
Why does this matter more than ever right now? According to the Flexera 2025 State of the Cloud Report released in March 2025, 84% of organizations consider managing cloud spend their top cloud challenge today. And the problem scales with the market. Gartner projects that global end-user spending on public cloud services will reach $723.4 billion in 2025, up from $595.7 billion in 2024 — a 21.5% increase. The larger the spending base, the more money is at stake with every percentage point of inefficiency.
FinOps is not a project with a finish line. It is an operational capability that, once in place, continuously reduces waste and turns cloud cost from an accounting surprise into a business lever.
The good news is that this does not require rebuilding your platform. Most saving levers rest on a solid cloud engineering foundation and a well-executed migration; if that foundation is not solid, it is worth reinforcing it before optimizing costs. Our cloud and DevOps services team works precisely at that intersection of architecture, automation, and cost.
The three phases of the FinOps framework: Inform, Optimize, and Operate
The FinOps Foundation framework is structured around three iterative phases — not sequential steps you complete once. You cycle through them continuously, and they also follow a Crawl-Walk-Run maturity approach: you start with the basics (Crawl), extend them (Walk), and finally automate and optimize deeply (Run).
Inform: visibility, allocation, budgeting, and forecasting
You cannot optimize what you cannot see. The Inform phase is about achieving granular visibility into spend, correctly allocating it to teams, products, or business units (through well-structured tags and accounts), establishing budgets, and building reliable forecasts. This is where you set up your tagging system, define your cost taxonomy, and connect billing data to an analytics tool. Without a solid Inform phase, everything else is built on sand: you will optimize blindly and be unable to attribute savings to anyone.
Optimize: eliminating waste and driving efficiency
With real visibility, the Optimize phase attacks waste head-on: rightsizing over-provisioned instances, shutting down idle resources, committing to discount programs in exchange for reserved usage, and selecting the right purchasing model for each workload. It is no accident that this phase delivers the highest immediate return. According to the FinOps Foundation's State of FinOps 2025 Report, workload optimization and waste reduction is the number-one priority for FinOps practitioners for the second consecutive year, cited by 50% of respondents as their top focus.
Operate: KPIs, governance, and continuous improvement
The Operate phase turns one-off savings into a sustainable practice. This is where you define KPIs (cost per business unit, committed discount coverage percentage, waste identified vs. eliminated), establish governance (mandatory tagging policies, anomaly alerts, approval workflows), and run the continuous improvement cycle. Without Operate, the waste you eliminated in an optimization campaign creeps back within a few months because teams keep deploying with the same habits.
| Phase | Question it answers | Key activities | Typical KPI |
|---|---|---|---|
| Inform | How much are we spending, and on what? | Tagging, allocation, budgets, forecasting | % of spend correctly allocated |
| Optimize | Where can we save without losing value? | Rightsizing, committed discounts, resource cleanup | Waste eliminated, % discount coverage |
| Operate | How do we sustain this over time? | KPIs, governance, anomaly alerts, continuous improvement | Cost per business unit, budget variance |
How much cloud spend is actually wasted?
It is the question every CFO asks and almost no organization can answer precisely. The industry reference figure comes from the Flexera 2025 State of the Cloud Report: organizations are still wasting 27% of their cloud spend. That number has improved (it peaked at 32% four years earlier), suggesting FinOps practices are starting to pay off — but it is still enormous. On an annual bill of one million dollars, that is $270,000 that, in theory, buys no value.
That waste is not a moral failing of engineering teams — it is structural. It accumulates in very specific, repeatable patterns:
- Over-provisioned instances: machines with 16 vCPUs running at 8% CPU because someone "played it safe" at deployment.
- Orphaned resources: disks, IPs, snapshots, and load balancers that outlive the workload that justified them.
- Non-production environments left running 24/7: development, test, and staging environments running around the clock when they are only used during business hours.
- Absence of committed discounts: paying On-Demand prices for stable, predictable workloads that have been running the same way for two years.
- Lack of allocation: without tagging, nobody is accountable, so nobody cleans up.
The compounding factor is growth rate. The same Flexera report indicates that cloud spend is expected to grow 28% in the following year. If waste holds at 27% on a base growing at double digits, the absolute cost of that inefficiency increases every year even if the percentage stays flat. That is why spending management cannot be an annual effort — it needs to be a permanent operational capability.
Real saving levers on AWS, Azure, and GCP
This is where FinOps moves from theory into practice. The concrete levers fall into three categories: discount purchasing models, resource efficiency, and architecture. Here are the most powerful ones for each provider.
Purchasing models: the fastest savings
The first win usually comes from paying less for the same thing. On AWS, according to official Savings Plans and EC2 Spot Instances documentation, commitment-based options offer very significant discounts over On-Demand pricing:
| AWS purchasing model | Maximum discount vs. On-Demand | When to use it |
|---|---|---|
| EC2 Instance Savings Plans / Standard Reserved Instances | Up to 72% | Stable, predictable workloads within a specific instance family |
| Compute Savings Plans | Up to 66% | Stable workloads needing flexibility across families, regions, and services (EC2, Fargate, Lambda) |
| Spot Instances | Up to 90% | Interruption-tolerant workloads: batch, CI/CD, async processing, big data |
The same logic applies to the other providers under different names: Azure offers Reservations and Savings Plans for compute, plus Spot VMs; Google Cloud offers Committed Use Discounts (CUDs), Sustained Use Discounts, and Spot VMs. The principle is universal: commit stable usage in exchange for a discount, and reserve interruptible workloads for the cheapest spare capacity.
The operational key is not to commit blindly. First do rightsizing (so you are not reserving capacity you do not need), then analyze the stable usage pattern from recent months, and only then commit to a prudent coverage level while leaving room for organic growth.
Resource efficiency: waste you can see at a glance
- Rightsizing: adjust the size of instances, databases, and volumes to match observed real usage. This is the lever with the best effort-to-savings ratio.
- Shutdown scheduling: stopping development and test environments outside business hours can cut their cost by 65–70% without affecting production.
- Orphaned resource cleanup: delete unattached disks, old snapshots, unassigned IPs, and load balancers with no traffic.
- Storage tiering: move infrequently accessed data to cheaper storage classes (S3 Glacier, Azure Cool/Archive, GCS Nearline/Coldline) using automated lifecycle policies.
Architecture: structural, long-term savings
Architectural decisions deliver the largest long-term savings but require the most engineering investment. Moving from always-on instances to serverless models (Lambda, Azure Functions, Cloud Run) that scale to zero, adopting containers with real autoscaling, or consolidating underutilized databases, fundamentally transforms your cost structure. These decisions are best made during a migration or modernization project. If you are planning to move workloads to the cloud, it pays to do so with FinOps criteria from day one — something we address in our cloud migration service to avoid carrying on-premise inefficiencies into the cloud.
How to build a FinOps team and culture
The technology is the easy part. The real challenge is organizational: FinOps works when engineering, finance, and business share a common language and a shared responsibility. It is no coincidence that, to regain control of spending, organizations are increasingly turning to managed service providers (60%) and expanding their FinOps teams (59%), according to the Flexera 2025 State of the Cloud Report.
Who belongs on the team
A FinOps team is not an isolated department — it is a cross-functional capability. The typical roles are:
- FinOps practitioner / lead: orchestrates the practice, defines KPIs, and reports. The bridge between worlds.
- Engineering / platform / SRE: executes rightsizing, shutdown automation, and architectural decisions.
- Finance / FP&A: brings budgets, forecasting, and the connection to the company's financial planning process.
- Business / product: defines which workloads generate value and, therefore, where it makes sense to invest more or less.
The culture: making cost visible where decisions are made
The decisive cultural shift is moving cost information to the moment and place where technical decisions are made. When a product team can see on their dashboard what their service costs per active customer, the conversations change. Governance helps (mandatory tagging policies, anomaly alerts, monthly per-team cost reviews), but the real engine is distributed ownership: each team owns its bill and understands it.
A pattern that works is a regular cadence: a monthly cost review by business unit, an automated weekly anomaly report, and a quarterly optimization session where engineering and finance jointly prioritize the next set of levers. Without this rhythm, FinOps dissolves into good intentions.
First steps: from Crawl to Run in 90 days
Adopting FinOps does not require a year-long transformation. Following the FinOps Foundation's Crawl-Walk-Run maturity model, a realistic 90-day plan might look like this:
Days 1–30 (Crawl): see
- Activate your provider's native cost tool (AWS Cost Explorer, Azure Cost Management, GCP Cloud Billing) and, if needed, a cross-cloud analytics layer.
- Define and apply a minimum tagging taxonomy (environment, team, product, cost center).
- Identify the top 10 services by spend and obvious waste (orphaned resources, non-prod environments always on).
Days 31–60 (Walk): save
- Execute the first rightsizing cycle on your largest workloads.
- Schedule shutdown of non-production environments outside business hours.
- Analyze the stable usage pattern from recent months and purchase a first, conservative batch of committed discounts (Savings Plans, Reservations, or CUDs).
Days 61–90 (Run): sustain
- Establish KPIs and a cost-per-business-unit dashboard.
- Configure spending anomaly alerts and mandatory tagging policies.
- Introduce the review cadence (monthly cost review, weekly anomaly report, quarterly optimization session).
After one quarter you will have real visibility, documented tangible savings, and a system that prevents that waste from accumulating again. From there, improvement is incremental and continuous.
Conclusion
FinOps is not a tool or a cost-cutting campaign — it is an operational capability that aligns engineering, finance, and business to maximize the value of every dollar invested in the cloud. With 27% of cloud spend still wasted and budgets overrunning by an average of 17% according to Flexera, the saving potential is real and measurable. The levers exist (committed discounts of up to 72% on AWS, rightsizing, environment shutdowns, serverless architecture) and the Inform-Optimize-Operate framework offers a proven path to capture them and, above all, to hold on to them.
If you want to turn your AWS, Azure, or GCP cloud bill into a controlled business lever, at Technova Partners we help companies build their FinOps practice, execute the first optimization cycle, and sustain savings over time. Talk to us about your situation and let's design your 90-day roadmap together.
