The Cloud Cost Problem Is Getting Worse
Cloud spending continues to grow at double-digit rates for most enterprises, but the value delivered per dollar spent is not keeping pace. The causes are well-understood: over-provisioned resources, idle infrastructure, lack of visibility into who is spending what, and engineering teams that optimise for speed and reliability without considering cost. The result is cloud bills that surprise finance teams and erode the business case for cloud investment.
What FinOps Actually Is
FinOps is frequently mischaracterised as a cost-cutting exercise. It is not. It is a cultural and operational practice that brings financial accountability to cloud spending — enabling organisations to make informed trade-offs between speed, quality, and cost. The FinOps Foundation defines three phases: Inform (visibility into spending), Optimise (acting on that visibility), and Operate (embedding cost awareness into ongoing engineering practice). Most organisations that struggle with cloud costs are stuck in the Inform phase — they have dashboards but not decisions.
The Tagging Discipline Problem
Consistent resource tagging is the foundation of cloud cost visibility, and it is consistently the area where organisations fall shortest. Without reliable tags mapping resources to teams, applications, environments, and cost centres, it is impossible to allocate costs accurately or hold teams accountable for their spending. We recommend treating tagging as a compliance requirement — enforced through policy-as-code, with automated remediation for non-compliant resources and regular audits of tag coverage and accuracy.
Reserved Instances and Savings Plans: Getting the Balance Right
Commitment-based discounts — Reserved Instances and Savings Plans on AWS, Reserved VM Instances on Azure — can deliver 30–60% savings on compute costs compared to on-demand pricing. But the commitment creates risk: over-commit and you pay for capacity you do not use; under-commit and you leave savings on the table. The organisations achieving the best results use a portfolio approach: a base layer of 1-year commitments covering predictable workloads, a smaller layer of 3-year commitments for stable, long-lived infrastructure, and on-demand or spot capacity for variable and fault-tolerant workloads.
Rightsizing: The Ongoing Practice
Rightsizing — matching resource allocation to actual utilisation — is not a one-time exercise. Workload patterns change, applications evolve, and the cloud provider's instance catalogue expands regularly. Organisations that treat rightsizing as a quarterly or annual exercise consistently find that savings erode between reviews. The most effective approach is continuous rightsizing: automated recommendations surfaced to engineering teams as part of their regular workflow, with clear ownership and accountability for acting on them.
Building a FinOps Culture
The technical practices of FinOps — tagging, rightsizing, commitment management — are relatively straightforward to implement. The harder challenge is cultural: getting engineering teams to care about cost as a first-class engineering concern alongside performance and reliability. The organisations that succeed at this treat cost as a product metric, surface cost data in the same dashboards as performance and availability metrics, and celebrate cost efficiency as a form of engineering excellence rather than a constraint imposed by finance.