Cloud waste isn’t a rounding error anymore it’s a structural cost of doing business. Flexera’s 2025 State of the Cloud Report found that organizations waste an average of 27% of their cloud spend, and 84% of organizations now call cloud cost management their top cloud challenge. The figure has barely moved in five years, which tells you something important: most organizations aren’t short on cost-management tools. They’re short on a methodology that optimizes spend without engineering teams treating it as a threat to their velocity.

That tension is exactly what one global enterprise a multi-region AWS environment supporting hundreds of microservices across several business units was navigating when Lean IT was brought in. Their instinct, like most engineering organizations, was to resist cost initiatives outright: prior attempts at “cost cutting” had meant blanket instance downsizing that caused latency regressions, which made every subsequent optimization conversation an uphill argument with the platform teams who’d own the consequences.

The approach that worked was technical, not political. Lean IT’s engagement started with workload-level cost attribution tagging and allocating spend by service and team rather than relying on account-level totals, which is where most FinOps efforts stall before they generate trust. From there, the optimization split into three concurrent workstreams: rightsizing based on actual CPU/memory utilization data rather than peak-provisioned capacity, migrating steady-state, predictable workloads to Reserved Instances and Savings Plans while keeping genuinely variable workloads on-demand or on Spot, and restructuring storage tiering so infrequently accessed data moved out of S3 Standard automatically via lifecycle policies instead of manual review.

None of these levers required re-architecting applications, and that distinction mattered. Engineering teams kept shipping on their existing roadmaps while the cost workstreams ran in parallel the optimization was infrastructure-layer, not application-layer, so release velocity wasn’t a tradeoff being negotiated against savings. Over the engagement, the organization reduced its AWS bill by 32%, with the majority of savings coming from rightsizing and commitment-based discounting rather than workload reduction meaning the same compute capacity, used more efficiently, not less compute available to the business.

The pattern holds beyond this one engagement. Flexera’s research shows idle compute and over-provisioned instances are consistently the two largest categories of cloud waste industry-wide, which is precisely where a structured, data-driven FinOps review tends to find the fastest, lowest-risk savings before anyone has to touch application architecture at all.

This is the discipline Lean IT brings to cloud cost engagements: treating FinOps as an engineering problem with measurable inputs, not a procurement exercise layered on top of infrastructure teams. Cost attribution first, optimization second, governance third so savings persist instead of drifting back to baseline within two quarters, which is what happens when cost reviews are one-time events instead of an operating model.

If your AWS bill keeps climbing and nobody can confidently say which team or workload is driving it, that’s the first problem to solve not the last.

Schedule a consultation call with Lean IT to find out where your AWS spend is actually going, and how much of it can come back without slowing your engineering roadmap.