Cloud bills rarely grow because of one bad decision. They grow through many small ones: a test environment nobody switched off, a database sized for a launch that never came, logs kept forever. The good news is that the same few checks find most of the waste.
Visibility first
- Tag every resource with an owner, environment and product
- Set budgets and alerts per team, not just per account
- Review a cost dashboard in the same meeting as your product metrics
The usual suspects
- Idle or oversized virtual machines and databases: rightsize based on real usage
- Non-production environments running nights and weekends: schedule them off
- Unattached disks, old snapshots and orphaned load balancers
- Storage that never moves to cheaper tiers
- Data transfer between regions and zones that could be avoided
- Logs and metrics kept far longer than anyone reads them
Commitments, once the baseline is clean
Reserved instances and savings plans can cut steady-state compute costs substantially, but only commit after rightsizing, or you lock in the waste.
Make it a habit
Cost optimisation is not a one-off project. A monthly review with engineering and finance, and cost checks in your infrastructure-as-code reviews, keep savings from creeping back.
Our engineers and consultants write about what they see in real projects. Have a question about your own systems?






