The FinOps Framework Explained
Cloud computing promised to reduce IT costs through pay-as-you-go flexibility. For many enterprises, it delivered the opposite: spiralling bills, budget overruns, and shadow spending that finance teams struggle to track. The FinOps Foundation, which governs the FinOps Framework, estimates that organisations practicing mature FinOps reduce cloud waste by an average of 30% compared to those without a structured programme.
FinOps is not a tool or a product — it is a cultural and operational practice that connects three functions: Engineering (who create the spend), Finance (who budget and account for it), and Business (who derive value from it). The framework operates in three iterative phases: Inform (gaining visibility into what you're spending and why), Optimise (taking action to reduce waste and maximise discount utilisation), and Operate (building governance processes that prevent future waste).
FinOps Maturity Signal
Organisations at a "Crawl" FinOps maturity level typically have unit economics visibility for 30% of their estate. "Run" organisations have 90%+ coverage with automated anomaly detection and real-time cost allocation to business units. Most enterprises start at Crawl and reach Walk within 12–18 months of a structured programme.
The FinOps Framework identifies six domains: Understanding Cloud Usage and Cost, Performance Tracking and Benchmarking, Real-Time Decision Making, Cloud Rate Optimisation, Cloud Usage Optimisation, and Organisational Alignment. This guide covers all six through the lens of enterprise procurement and negotiation strategy — particularly how to use commercial leverage to achieve the discounts and terms that FinOps tools alone cannot deliver.
For a deep dive into building a FinOps culture, see our guide to FinOps for Enterprises: Building a Cloud Cost Culture.
Understanding Cloud Waste: Where the Money Goes
Before any optimisation effort, you need a clear picture of your waste profile. Cloud waste falls into five broad categories, each with distinct remediation strategies and savings potential.
Expert Advisory
Want independent help negotiating better terms? We rank the top advisory firms across 14 vendor categories — free matching, no commitment.
| Waste Category |
Typical % of Spend |
Time to Remediate |
Effort |
| Idle resources (stopped VMs, unattached storage) | 8–15% | Days–weeks | Low |
| Over-provisioned instances (size too large) | 10–20% | Weeks–months | Medium |
| Underused commitment purchases (RI/SP coverage gaps) | 5–12% | 1–3 months | Medium |
| Non-production environments running 24/7 | 8–18% | Weeks | Low–medium |
| Architectural inefficiency (wrong service, over-engineered) | 5–15% | Months–quarters | High |
The first two categories — idle resources and over-provisioned instances — are the fastest wins and should be the focus of any organisation in the first 90 days of a FinOps programme. Cloud provider native tools (AWS Cost Explorer, Azure Advisor, GCP Recommender) surface these opportunities automatically; the challenge is building the organisational process to act on recommendations consistently.
Non-production environments are a particularly high-value target. Dev/test/staging environments often mirror production in size, yet run continuously even when developers are not working. Implementing automated scheduling to shut down non-production workloads outside business hours typically saves 30–60% of non-production compute costs — often 10–18% of total cloud spend.
Our dedicated guide to Cloud Waste: How Enterprises Lose 30% of Their Cloud Spend walks through elimination tactics in detail.
Warning: Egress Costs Are Often Invisible
Data transfer (egress) costs are consistently the most under-anticipated line item in enterprise cloud budgets. Moving data out of AWS, Azure, or GCP to the internet or to another cloud can cost $0.08–$0.09/GB — adding up to millions per year for data-intensive workloads. Review your cloud egress costs before committing to a multi-year spend agreement, as negotiated EDPs/MACCs rarely cover egress at preferential rates.
Commitment Strategy: Reserved Instances, Savings Plans, and CUDs
The single highest-impact lever for reducing cloud costs is purchasing committed-use discounts — Reserved Instances (RIs), Compute Savings Plans, EC2 Instance Savings Plans, or Google Cloud Committed Use Discounts (CUDs). Enterprises that systematically cover 60–80% of their baseline compute workloads with commitments can reduce effective compute rates by 30–60% compared to on-demand pricing.
AWS Commitment Options
AWS offers three tiers of commitment discount:
- Reserved Instances (RIs): Commit to a specific instance family, region, OS, and tenancy for 1 or 3 years. Standard RIs offer up to 72% savings; Convertible RIs allow instance type changes in exchange for slightly lower discounts (up to 66%). RIs are still the best choice for RDS, ElastiCache, Redshift, and OpenSearch where Savings Plans don't apply.
- Compute Savings Plans: Commit to a consistent $/hour spend, with discounts automatically applied across EC2, Lambda, and Fargate regardless of instance type or region. Offers up to 66% vs on-demand and is generally preferred for EC2 due to flexibility.
- EC2 Instance Savings Plans: Commit to a specific instance family in a specific region for maximum discount (up to 72%), but with more flexibility than RIs on size and OS within that family.
For detailed AWS commitment strategy, see our AWS Cost Optimization guide and the dedicated comparison at Reserved Instances vs Savings Plans.
Azure Commitment Options
Azure Reserved VM Instances (RIs) offer 1 or 3-year commitments with discounts of 20–60% depending on instance type. Azure Hybrid Benefit (AHB) allows customers to apply existing Windows Server or SQL Server licences to Azure VMs, saving an additional 30–49% on eligible workloads. The combination of AHB and Azure RIs can reduce certain workload costs by 70%+ compared to pay-as-you-go. Full analysis in our Azure Cost Management guide.
Google Cloud CUDs
GCP Committed Use Discounts offer 1 or 3-year commitments on compute resources (vCPU and memory) with discounts of 37% (1-year) to 55% (3-year) for most machine types. GCP also applies Sustained Use Discounts (SUDs) automatically for workloads running more than 25% of the month — these require no commitment and provide up to 30% savings. The interaction between CUDs, SUDs, and GCP's Spot VM pricing creates a nuanced optimisation problem. See our GCP cost optimisation guide for the full picture.
Commitment Coverage Target
Industry benchmark: enterprises with mature FinOps programmes target 70–80% commitment coverage of their steady-state baseline workloads. The remaining 20–30% is covered by on-demand or Spot/Preemptible instances for variable and ephemeral workloads. Going above 80% increases risk of over-commitment if workloads are decommissioned — commitments are not refundable unless resold through the marketplace.
AWS, Azure, and GCP: Per-Platform Tactics
While the FinOps framework applies across all cloud providers, each hyperscaler has unique pricing mechanics, discount programmes, and negotiation dynamics that require platform-specific expertise.
Free Resource
Get the IT Negotiation Playbook — free
Used by 4,200+ IT directors and procurement leads. Oracle, Microsoft, SAP, Cloud — all covered.
AWS Cost Optimisation
AWS remains the largest cloud provider, and its pricing model is the most complex — with over 200 services, multiple commitment tiers, regional price variations, and a constantly evolving instance type catalogue. Key AWS-specific tactics include:
- Spot Instance strategy: Spot instances offer 60–90% savings for fault-tolerant, interruptible workloads (batch processing, CI/CD, data analytics). AWS Spot Fleet and EC2 Auto Scaling Groups can maintain availability across multiple instance types and Availability Zones to minimise interruptions. See our Spot Instance strategy guide.
- S3 Intelligent-Tiering: Automatically moves objects to lower-cost storage tiers based on access patterns, with no retrieval fees. Appropriate for data lakes and backup archives where access patterns are unpredictable.
- AWS Compute Optimizer: Machine learning-based recommendations for EC2, EBS, Lambda, and ECS rightsizing. Free to use; typically identifies 15–30% of instances as over-provisioned.
- Graviton migration: AWS Graviton (ARM-based) instances offer 20–40% better price-performance than equivalent x86 instances. Workloads that have been containerised or are cloud-native typically migrate with minimal code changes.
Full tactics across 20 strategies: AWS Cost Optimization: 20 Strategies That Save Millions.
Azure Cost Optimisation
Azure's strengths are its integration with Microsoft's software estate (Windows, SQL Server, Teams, Microsoft 365) and its hybrid benefit licensing programme. The biggest Azure-specific opportunities are:
- Azure Hybrid Benefit (AHB): Applying existing on-premises SQL Server and Windows Server licences to Azure VMs. Frequently under-utilised — many enterprises have the licences but haven't activated AHB. Detailed analysis in our Azure Hybrid Benefit guide.
- Azure Dev/Test pricing: Significant discounts for development and test workloads through MSDN subscriptions. Can reduce dev environment costs by 40–60%.
- Azure Cost Management + Power BI: Native cost management tooling with tag-based cost allocation and budget alerts. Less powerful than third-party FinOps platforms but sufficient for organisations early in their FinOps journey.
Full Azure playbook: Azure Cost Management: Enterprise Playbook.
Google Cloud Cost Optimisation
GCP's differentiator is its automatic discounting model — SUDs, preemptible VMs, and per-second billing that rewards dynamic workloads. Key GCP-specific tactics include leveraging BigQuery flat-rate pricing for predictable analytics spend, using GKE Autopilot for containerised workloads to eliminate node management overhead, and negotiating Google's CUDS on Compute Engine alongside commitments on BigQuery and Cloud SQL. See our GCP Cost Optimisation: CUD vs SUD Strategy guide.
Enterprise Discount Programme Negotiation
For enterprises spending $1M+ annually on cloud, negotiated enterprise agreements represent one of the highest-leverage cost reduction opportunities available. AWS EDP, Azure MACC, and GCP Committed Use Agreements all provide spend commitments in exchange for negotiated discount rates — but the headline discount is just the starting point.
Negotiation Leverage Points
The most effective negotiation leverage comes from: (1) multi-cloud competition — making AWS compete with Azure and GCP for your commitment; (2) migration commitments — offering to migrate on-premises workloads or competitor cloud workloads in exchange for credits and discounts; (3) commitment size — larger total commitments unlock deeper discount tiers; (4) contract duration — 3-year commitments typically unlock 5–10% more discount than 1-year deals.
AWS EDP agreements typically provide 10–25% discount on all eligible spend (excluding support, marketplace third-party, and some managed services) in exchange for committing to a minimum annual spend with penalty provisions for shortfall. The key negotiation variables are: discount rate, eligible services, ramp schedule (graduated commits in year 1), shortfall provisions, and marketplace credits.
Azure MACC (Microsoft Azure Consumption Commitment) provides a committed spend pool that can be drawn down against Azure services. MACCs interact with broader Microsoft EA negotiations — enterprises renewing Microsoft EA should negotiate MACC terms simultaneously to maximise combined leverage. See our guide to Azure Committed Spend Negotiation for the full playbook.
For a side-by-side comparison of all three providers' enterprise programmes, see our guide to Negotiating Enterprise Discount Programs: EDP, MACC, and CUD Agreements.
Cloud Cost Governance Framework
Optimisation without governance is a temporary fix. Without structural controls, cloud costs creep back up as new teams spin up resources, commitments expire without renewal, and architectural decisions are made without cost accountability. A mature cloud cost governance framework addresses four pillars:
| Governance Pillar |
Key Mechanisms |
Responsible Party |
| Visibility & Allocation | Tag policy enforcement, showback/chargeback model, business unit reporting | FinOps team + Engineering |
| Budget & Forecasting | Budget alerts, anomaly detection, spend forecasting models | Finance + FinOps |
| Procurement Controls | Commitment approval workflows, reserved capacity management, EDP/MACC governance | Procurement + FinOps |
| Engineering Standards | Cloud cost guardrails in CI/CD, architecture review boards, resource sizing standards | Engineering + Platform |
Tagging is the foundation of cloud cost governance. Without consistent resource tagging by environment, application, cost centre, and team, cost allocation becomes guesswork and accountability breaks down. Establishing a mandatory tagging policy — enforced via cloud policy engines (AWS Service Control Policies, Azure Policy, GCP Organisation Policies) and automated remediation for non-compliant resources — is a prerequisite for any meaningful FinOps programme.
Full governance framework: Cloud Cost Governance: Policies Every Enterprise Needs. For allocation methodology, see Building a Cloud Cost Allocation Strategy.
The FinOps tools market has matured rapidly, with solutions ranging from cloud-native cost management consoles to sophisticated third-party platforms offering AI-driven recommendations, multi-cloud normalisation, and showback automation. Understanding where each category adds value — and where it doesn't — is essential for selecting the right tooling for your organisation.
Cloud-native tools (AWS Cost Explorer, Azure Cost Management, GCP Billing) are free and provide the most accurate data for their respective platforms. They are sufficient for single-cloud organisations and for initial visibility work. Their limitation is the inability to provide normalised multi-cloud views or automated optimisation actions across providers.
Third-party FinOps platforms (Apptio Cloudability, CloudHealth, Spot.io, Harness Cloud Cost Management, CloudZero) add multi-cloud normalisation, unit economics dashboards, commitment purchase automation, and showback/chargeback capabilities. They typically cost 1–3% of managed cloud spend, and ROI is straightforward to demonstrate when savings exceed the platform cost. See our full comparison at FinOps Tools Comparison 2026: Which Platform Is Best?
Need help selecting a FinOps platform or negotiating cloud enterprise agreements?
Independent advisors with 500+ cloud engagements — no vendor conflicts
Get Matched →
10 Cloud Cost Optimization Tactics for Enterprise Teams
Tactic 01
Establish a FinOps Function Before Optimising
Sustainable cloud cost reduction requires organisational change, not just tooling. Designate a FinOps lead (or team in larger enterprises), establish cross-functional working groups with Engineering and Finance, and define shared KPIs before embarking on technical optimisation. Without this foundation, savings achieved by one initiative are often offset by new spending from teams with no cost accountability.
Tactic 02
Run a 30-Day Waste Audit Before Any Commitment Purchases
Buying Reserved Instances or Savings Plans before rightsizing your workloads locks in over-provisioned capacity. Always complete a waste audit first: identify idle resources, stop or terminate them, downsize over-provisioned instances, and implement non-production scheduling. Only then should you size and purchase commitments against the optimised baseline — otherwise you're committing to pay for waste for 1–3 years.
Tactic 03
Target 70–80% Commitment Coverage on Steady-State Workloads
Compute Savings Plans and Reserved Instances should cover your stable, predictable baseline. Analyse 3–6 months of usage data to identify consistent on-demand spend that can be converted to committed pricing. For most enterprises, this means covering production databases (RIs), core application servers (Savings Plans), and persistent platform services (Cloud SQL CUDs). Leave dynamic, autoscaling, and batch workloads on Spot/on-demand.
Tactic 04
Negotiate EDP/MACC with Multi-Cloud Competition
AWS, Azure, and GCP all compete aggressively for committed spend. Running a competitive process — even if you have a strategic preference — forces providers to sharpen their commercial terms. Enterprise teams have achieved 5–10% additional discount headroom by demonstrating credible alternatives. Frame your RFP around total committed spend, support requirements, marketplace credits, and migration assistance, not just headline discount rates.
Tactic 05
Implement Storage Lifecycle Policies Immediately
Object storage (S3, Azure Blob, GCS) accumulates silently and costs accumulate proportionally. Data that was needed for an active project often sits in hot storage tiers for months or years after the project ends. Implementing lifecycle policies to transition objects to infrequent access tiers after 30–90 days, and to archive tiers after 180 days, can reduce storage costs by 40–70% with zero performance impact on active workloads. See our
Cloud Storage Optimisation guide.
Tactic 06
Enforce Tagging Policy with Automated Remediation
Untagged resources are invisible to cost allocation and immune to accountability. Implement cloud policy guardrails that prevent resource creation without mandatory tags (environment, application, team, cost centre) and that automatically flag or quarantine untagged resources created via the console or APIs. Tools like AWS Config Rules, Azure Policy, and GCP Organisation Policies make this enforceable without manual audit overhead.
Tactic 07
Use Spot/Preemptible Instances for Batch and CI/CD
Continuous integration pipelines, data processing jobs, rendering workloads, and machine learning training runs are ideal Spot candidates — they are interruptible and can be architected to checkpoint and restart. At 60–90% discount vs on-demand, moving even a modest volume of batch compute to Spot can save $100K+ annually for larger enterprises. See our
Spot Instance Strategy guide for interruption handling patterns.
Tactic 08
Review Cloud Architecture Against Cost-Optimised Patterns
Some cloud costs are structural — driven by architectural decisions that create inefficiency regardless of configuration. Common patterns include monolithic applications that can't scale down, synchronous APIs that hold expensive compute waiting for I/O, and data pipelines that pull entire datasets when incremental processing would suffice. Architecture reviews focused on cloud cost efficiency typically identify 15–25% of spend as addressable through refactoring, but require cross-functional investment.
Tactic 09
Build Unit Economics into Engineering Decision Making
The most mature FinOps teams have embedded cost per unit (cost per transaction, cost per customer, cost per GB processed) into engineering metrics dashboards. When engineers can see in real time how their design decisions affect cost per unit, architectural improvements become self-directing. This requires investment in instrumentation and tooling, but creates a sustainable cost culture that doesn't depend on central review processes.
Tactic 10
Engage an Independent Advisor for EDP/MACC Negotiations
Cloud providers negotiate dozens of enterprise agreements per year. Most enterprise procurement teams negotiate one every 3 years. This knowledge gap is a structural disadvantage. An independent cloud negotiation advisor who has benchmarked recent EDP and MACC outcomes can tell you what discount rate is achievable for your spend level, which contractual protections to push for, and how to structure ramp schedules that protect you from shortfall penalties. See our
rankings of the best cloud negotiation consulting firms to find qualified advisors.
Deep-Dive Guides in This Series
This pillar guide covers the full cloud cost optimisation landscape. For deep dives into specific topics, platforms, or strategies, explore the complete Cloud Cost & FinOps series:
FinOps Culture
FinOps for Enterprises: Building a Cloud Cost Culture
How to stand up a FinOps function, drive cross-team engagement, and build sustainable cost governance inside large organisations.
AWS
AWS Cost Optimization: 20 Strategies That Save Millions
The complete AWS playbook: Savings Plans, Spot, Graviton, RDS optimisation, S3 lifecycle, and EDP negotiation tactics.
Azure
Azure Cost Management: Enterprise Playbook
Azure Reserved Instances, Hybrid Benefit, Dev/Test pricing, MACC negotiation, and Cost Management tooling guide.
Google Cloud
GCP Cost Optimization: CUD vs SUD Strategy
When to use Committed Use Discounts vs Sustained Use Discounts, Spot VMs, BigQuery pricing, and GCP CUD negotiation.
Multi-Cloud
Multi-Cloud Cost Optimization: Managing AWS + Azure + GCP
Normalising costs across providers, multi-cloud FinOps tooling, and avoiding commitment conflicts in multi-cloud estates.
Commitment Strategy
Reserved Instances vs Savings Plans: Complete Comparison
When to use RIs vs Savings Plans on AWS, and equivalent commitment programmes on Azure and GCP.
Waste Elimination
Cloud Waste: How Enterprises Lose 30% of Their Cloud Spend
The five waste categories, how to identify them, and the remediation playbook for each — with quick-win prioritisation.
Negotiation
How to Negotiate Enterprise Discount Programs (EDP/MACC)
EDP vs MACC vs GCP CUD agreement structures, benchmark discount rates, and negotiation tactics for large enterprises.
Tooling
FinOps Tools Comparison 2026: Which Platform Is Best?
Side-by-side comparison of Apptio, CloudHealth, Spot.io, Harness, CloudZero, and native tools. With selection criteria.
Compute
Spot Instance Strategy: Cutting Compute Costs by 70%
How to architect for Spot interruptions, which workloads qualify, and how to use Spot Fleet and Auto Scaling Groups.
Governance
Cloud Cost Governance: Policies Every Enterprise Needs
Tagging policies, budget guardrails, commitment approval workflows, and the governance operating model for cloud FinOps.
Allocation
How to Build a Cloud Cost Allocation Strategy
Showback vs chargeback models, shared cost allocation methods, tagging strategies, and business unit reporting frameworks.
Frequently Asked Questions
What is FinOps and why does it matter for cloud cost optimization?
FinOps (Financial Operations) is a cloud financial management practice that brings together finance, engineering, and business teams to collaborate on cloud spending. It matters because enterprise cloud bills have become the fastest-growing technology cost line item — often exceeding forecasts by 30–40% — and because cloud pricing complexity makes optimisation impossible without a structured framework. FinOps creates shared accountability between the teams that consume cloud resources and the teams that pay for them.
How much can enterprises realistically save on cloud costs?
Industry data consistently shows that enterprises waste 20–35% of cloud spend on idle or over-provisioned resources. With a structured FinOps programme combining commitment purchases, right-sizing, and waste elimination, most large enterprises can reduce effective cloud spend by 30–50% compared to unmanaged on-demand consumption. Negotiated enterprise discount programmes (EDP, MACC, CUD agreements) can deliver an additional 10–25% discount on top of published rates for enterprises committing $1M+ annually.
What is the difference between Reserved Instances and Savings Plans?
Reserved Instances commit to a specific instance type, region, and OS for 1 or 3 years, achieving discounts of 30–72% vs on-demand. AWS Savings Plans are a more flexible commitment model where you commit to a consistent $/hour spend, and discounts apply automatically across EC2, Lambda, and Fargate regardless of instance type or region. Savings Plans are generally preferred for EC2 modern workloads due to flexibility; RIs remain better for database services (RDS, ElastiCache) where Savings Plans don't apply.
When should an enterprise negotiate an AWS EDP or Azure MACC?
EDP and MACC agreements become viable at $1M+ annual cloud spend. The optimal negotiation window is 3–6 months before your current contract or enterprise agreement expires. Running a competitive process across providers — even if you have a platform preference — is the most effective way to achieve maximum discount. Engaging an independent advisor who can benchmark recent deal outcomes is strongly recommended; most internal procurement teams lack the frequency of negotiation to assess what is achievable.
What cloud cost optimisation quick wins should enterprises tackle first?
The highest-ROI quick wins are: (1) eliminating idle resources — stopped VMs, unattached storage, unused load balancers — typically saving 8–15% in weeks; (2) purchasing Savings Plans or RIs for steady-state baseline workloads after rightsizing; (3) rightsizing over-provisioned instances using cloud provider recommendations; (4) implementing storage lifecycle policies to move infrequently accessed data to cheaper tiers. These four actions alone can deliver 25–35% savings before any architectural changes are needed.