How it took us just 12 months to save $800k in AWS costs

Rather than implementing these approaches sequentially, we executed them in parallel across multiple business units.

Each business unit had dedicated champions driving dashboards, rightsizing, idle resource cleanup, and savings plan optimisation simultaneously, working closely with our platform team for enablement and support. This collaborative, distributed approach accelerated our progress and ensured buy-in from teams who felt ownership of their optimisation efforts.

We implemented CloudZero to provide real-time cost visibility across all AWS accounts and services. The key to making this work was our tagging strategy: we encouraged teams to tag their resources first, while we tagged accounts at the organisational level. In CloudZero, we layer resource and account-level tags, allowing for multitenant account flexibility.

This approach allowed each business unit to view its costs clearly while maintaining the flexibility to segment data further based on its own specific needs. Crucially, when resources lacked tags, the account owner became the allocated payer, ensuring that no costs fell through the cracks. This created a natural incentive for business units to correct allocations and maintain proper tagging practices.

The transformation was immediate. For the first time, we had complete transparency into cloud spending, enabling data-driven decision-making across all business units. Teams could finally see the cost impact of their infrastructure choices, and our finance team could accurately allocate cloud costs back to the responsible business units. This visibility didn't just improve reporting – it fundamentally changed behaviour.

We took a systematic, data-driven approach to rightsizing. Our comprehensive audit of all AWS services and utilisation patterns revealed significant opportunities for optimisation. We implemented automated monitoring to track resource usage over time. Then we began gradual rightsizing with careful performance validation.

Key wins:

Result: 4% cost reduction while maintaining performance standards.

We implemented automated scheduling for non-production environments, ensuring that development and testing resources ran only when needed. Auto-scaling policies based on actual demand replaced static provisioning, while regular cleanup procedures removed unused snapshots, volumes, and other resources.

Major cleanups:

Result: 5% reduction in idle resource costs, particularly in development and testing.

We analysed historical usage patterns thoroughly to understand our actual resource consumption. This analysis revealed opportunities for significant savings through a strategic mix of Compute and EC2 Instance Savings Plans. We also established regular review and adjustment processes to ensure our commitments matched our evolving needs.

Notable implementations:

Result: 13% additional savings on committed workloads while maintaining flexibility.

The numbers tell the story: $800,000 saved in 12 months with no impact on our service performance.

Notable wins:

Achieved Beyond the immediate savings, we gained operational excellence, gaining better visibility into cloud spending, faster decision-making with real-time cost data, and enhanced collaboration between engineering and finance teams.

Start with culture, not technology
We spoke to all teams to get buy-in for the FinOps project and explained why it was important to the business as a whole. We provided training and communicated with them extensively before we acted. This cultural foundation proved essential when we needed buy-in for changes that might initially seem inconvenient.

Make data-driven decisions
Every optimisation was backed by a comprehensive analysis. We implemented detailed monitoring before making changes, ensuring we had baseline data to measure against. This approach prevented us from making changes that looked good on paper but had unintended consequences in practice.

Iterate gradually
Rather than sweeping changes, we validated each optimisation before moving to the next. This cautious approach might seem slow, but it prevented costly mistakes and built confidence across the organisation. Teams were more willing to embrace changes when they could see the positive results from previous optimisations.

Monitor continuously
Cost optimisation is an ongoing discipline, not a one-time project. We established regular review cycles and continuous monitoring to ensure our optimisations remained effective as our infrastructure evolved. This discipline has become part of our operational DNA.