Cloud Cost Optimisation: Cutting AWS Bills Without Cutting Corners

Cloud bills have a way of becoming one of the largest line items in a growing company's infrastructure budget before anyone quite notices how it happened. We've conducted cost optimisation audits for dozens of companies and the same patterns appear with remarkable consistency. The good news: in nearly every case, 30-50% savings are achievable without meaningfully changing the reliability or performance of the system.

The usual culprits

Right-sizing is the single biggest opportunity in almost every audit. EC2 instances that were sized for a peak load that never materialised, RDS instances running at 8% average CPU, Elastic Load Balancers with almost no traffic. The starting point is always a utilisation report across every major service — you can't optimise what you haven't measured.

The optimisation playbook

• Right-size all compute and database instances based on 90-day utilisation data
• Convert on-demand instances to Reserved or Savings Plans for steady-state workloads
• Move infrequently accessed S3 data to Intelligent-Tiering or Glacier
• Eliminate idle resources — stopped instances still generating EBS charges are common
• Set up budget alerts and cost anomaly detection before making any changes
• Review data transfer costs — cross-AZ traffic is frequently overlooked and significant

The conversation about cutting corners

Every cost optimisation conversation eventually surfaces a proposal that sounds like a good idea until you think it through — running production databases on single-AZ, eliminating CloudWatch logging to save a few dollars, or using spot instances for workloads that can't tolerate interruption. Our rule is simple: we only recommend changes that a reasonable engineer would be comfortable defending in a post-incident review. Cost savings that introduce reliability risk are not savings — they are deferred incidents.

The cheapest cloud setup is the one that never pages your on-call engineer at 3am. Optimise cost. Never optimise away reliability.

The best time to run a cloud cost audit is right after a period of rapid growth — when infrastructure was provisioned fast and sized generously. The second-best time is now. A well-run audit typically pays for itself in the first month and delivers compounding savings for years after.