Step 1: Establish Visibility and Tagging
No optimization is possible without visibility. The first step—the Inform phase of FinOps—is to know precisely who spends what, where, and why. This requires collecting and consolidating cost data from all environments, ideally via the FOCUS standard if the organization is multi-cloud, in order to avoid rebuilding a data pipeline for each provider.
The nerve of the operation is tagging (resource labeling). Without a coherent tagging strategy—by team, project, environment, cost center—spending remains an indistinct mass that’s impossible to attribute. Defining a mandatory tag taxonomy and applying it from resource creation is the most cost-effective initial investment in a FinOps approach.
This visibility leads to shared dashboards accessible to both technical teams and finance. The goal isn’t to produce another monthly report, but to make cost visible on a daily basis, where decisions are made. An expense that nobody sees is an expense nobody optimizes.
Visibility also includes forecasting. Beyond recounting past expenditures, a mature FinOps approach anticipates future costs based on projects and expected growth. This projection helps establish credible budgets, detects potential drift early, and aligns finance and engineering teams around common expectations—avoiding end-of-month surprises and emergency last-minute arbitrage.
Step 2: Optimize through Rightsizing and Commitments
Once visibility is established, the Optimize phase targets concrete waste reduction. The most impactful lever for most organizations is rightsizing: adjusting resource sizes to actual usage. Providers consistently observe that the majority of virtual machines are oversized—you pay for capacity that is never used.
Recommendation tools (AWS Compute Optimizer, Azure Advisor, GCP Recommender) automatically identify these gaps. In addition, there is the elimination of orphaned resources: detached storage volumes, idle reserved IP addresses, obsolete snapshots, test environments left active.
For predictable workloads, price optimization completes the setup. Reserved instances and savings plans offer substantial reductions compared with on-demand pricing, in exchange for a time commitment. The trade-off is to cover the stable portion of consumption with commitments, while preserving on-demand flexibility for spikes.
A complementary lever worth mentioning: spot instances, cloud capacity offered at deep discounts (up to 90% off) but reclaimable by the provider at any moment. They are suitable for interruption-tolerant workloads—batch processing, test environments, certain compute workloads. When well orchestrated, especially via Kubernetes, they can significantly cut the bill for non-critical workloads. The combination of on-demand, commitments, and spot, tailored to each workload’s profile, sits at the heart of price optimization.
Step 3: Automate and Govern
Quick wins are useless if waste reappears just as quickly. The Operate phase anchors gains over the long term through automation and governance—this is what differentiates a mature FinOps approach from a simple sweep.
Automation takes several forms: scheduled shutdowns of development environments at night and on weekends, automatic deletion of untagged or orphaned resources, dynamic scaling based on load. These mechanisms eliminate waste without repeated human intervention.
A rising trend for 2025-2026 is agentic FinOps: platforms that no longer merely recommend but execute optimizations autonomously—rightsizing applied automatically, switching between spot and on-demand instances, ongoing optimization of Kubernetes clusters. This advanced automation requires robust guardrails and trust in the tool, as it acts directly on production infrastructure. It illustrates market direction: moving from visibility to automated action.
Policies and Guardrails
Governance sets the rules of the game: budgets by team, automatic alerts in case of drift or abnormal increases, quotas, approval required beyond certain thresholds. The challenge is to make waste hard to create by design, rather than correcting it after the fact. Well-designed guardrails empower accountability without stifling innovation.
Step 4: Build Culture and Measure
FinOps fails when it remains confined to one team or is perceived as a budget police. Its success depends on building a genuine culture where every stakeholder—engineer, product manager, leader—incorporates cost into their decisions. This requires training, transparency, and ideally recognizing the teams that contribute to optimization.
Integrating finance teams from the outset of projects is a key factor: their expertise helps identify opportunities and prevent waste upstream rather than spotting it on the bill. FinOps becomes a continuous dialogue between the technical and financial functions, rather than an imposed arbitrage.
Measure to Progress
As with any discipline of continuous improvement, FinOps is steered by indicators: coverage rate by commitments, share of tagged resources, amount of waste eliminated, cost per unit of business value (cost per transaction, per customer, per feature). These metrics demonstrate the value of the approach and guide subsequent efforts.
Implementing FinOps is therefore not a time-limited project but the establishment of a durable capability to spend wisely. Visibility, optimization, automation, and culture form a cycle that repeats and refines itself. When well executed, this approach turns cloud cost from a painful inevitability into a controllable lever that serves the company’s efficiency and innovation.
This content is published by Mentioned
