AWS Cost Optimization: Save 40% on Your Cloud Bill

Cloud cost reviews tend to go one of two ways: either nobody's looked at the bill in a year and the low-hanging fruit is enormous, or the client has already done the obvious cuts and needs someone to find what's left. Either way, the 40% figure is real — we've hit it on multiple engagements — but it's rarely one single change.

Start with what's simply unused

• Unattached EBS volumes and old snapshots — often 10–15% of spend on its own
• Idle load balancers and NAT gateways left over from decommissioned services
• Non-production environments left running 24/7 instead of scheduled off-hours
• Over-provisioned RDS instances sized for a peak load that never materialised

Right-size, then commit

Once you've cut what's genuinely unused, the next lever is right-sizing compute against real utilisation data — not the instance size someone picked during initial launch under time pressure. Only after right-sizing do Reserved Instances or Savings Plans make sense; committing to the wrong size just locks in the waste for a year.

Architecture-level savings

The largest single win is usually architectural: moving infrequently accessed data to cheaper storage tiers, adding a CDN in front of origin traffic that didn't need to hit compute at all, or replacing an always-on service with something event-driven that scales to zero. These take longer to implement than a tagging cleanup, but they're the ones that keep paying off month after month.

“Cost optimisation isn't a one-time project — it's a review cadence. The bill that's efficient today drifts within two quarters if nobody's watching it.”