A cloud cost optimization checklist you can run monthly

Von Rohit Athaide · Veröffentlicht am 3. Nov. 2026

Cost work goes best as a short, regular routine, not a big project once a year. This checklist is ordered from the checks that need the least effort and carry the least risk, to the ones that need more care. Run it monthly. Most months it will take under an hour.

Look first: where the money goes

1. Group last month’s cost by service, then by tag. The top three lines are where to look. Optimizing a line that is two percent of the bill is a poor use of time.
2. Compare with the month before. Anything that moved by more than ten percent deserves a reason.

Easy and safe

3. Delete idle resources. Stopped servers that still bill for storage, unattached disks, unused public addresses, old test environments.
4. Delete old snapshots and backups that are past your retention policy.
5. Set retention on logs. See logging and monitoring costs.
6. Turn off non-production at night and weekends where it can be scheduled. An environment that runs ten hours on weekdays only runs about 50 of the 168 hours in a week, roughly a third.

Moderate effort

7. Right-size servers. Compare CPU and memory use with what you pay for, and move to a smaller size where the headroom is large. See choosing a cloud VM size.
8. Check for newer instance generations and other architectures. Newer types often give more for the same price. See Arm versus x86.
9. Move cold data to cheaper storage classes, with lifecycle rules. See object storage classes and lifecycle.
10. Review data transfer. Find what leaves the cloud or crosses regions, and whether a cache or a different layout would reduce it. See using a CDN.

Needs care

11. Commit to the steady part of compute. Only after usage is stable. See reserved instances and savings plans.
12. Use spot for interruptible work. See spot and preemptible instances.
13. Review managed services against alternatives only when the line is large. Moving a database is risky and rarely worth it for a small saving.

Keep it from coming back

14. Tag everything and alert on untagged spend. See tags and labels.
15. Keep budgets and forecast alerts current. See budgets and cost alerts.
16. Write down what you changed and what it saved, so next month you can tell whether it held.

A caution

Saving money by removing resilience, backups or monitoring is a trade, not a saving. If a check would reduce your ability to recover from a failure, decide that on purpose, with the cost of an outage in mind. The list of common mistakes is in cloud cost mistakes small teams keep making, and the habit behind this routine is described in FinOps basics.