Cost & operations · Guide ·
How we run a monthly AWS cost review (framework you can steal)
A repeatable 90-minute review structure for finance and engineering — focused on decisions, not dashboards.
Cost reviews fail when they become screenshots of the billing console. This is the agenda we use with managed cloud clients — you can run a lighter version internally.
Before the meeting (async, 30 minutes)
- Pull last month’s spend by service and linked account (if using Organizations).
- Flag anything up >15% month-over-month without a known cause (launch, data load, new env).
- List top 5 resources by cost (often RDS, EC2, NAT, support, data transfer).
Send the summary to finance and engineering leads so the live session is about decisions.
Live session (60 minutes)
| Block | Focus |
|---|---|
| 0–10 min | Confirm anomalies — planned vs. surprise |
| 10–25 min | Top 3 workloads: right-sizing, reservations, architecture |
| 25–40 min | Waste hunt: idle resources, old snapshots, unused load balancers |
| 40–55 min | Assign owners and due dates (max 5 actions) |
| 55–60 min | Tagging / allocation gaps for next month |
After the meeting
- Document actions in a shared tracker (Notion, Jira, or a simple spreadsheet).
- Revisit open items first next month — consistency builds trust with finance.
What good looks like
Within two cycles, leadership should be able to answer: What did we spend? Why? What are we doing about it? without opening ten AWS consoles.
That clarity is often worth more than a single Reserved Instance purchase.