Your cloud bill doubled and nobody knows why: an SME cost-audit playbook

Cloud bills grow the way unused gym memberships do: slowly, invisibly, and with excellent explanations available for every line item. The four usual suspects, in order of size: forgotten environments (the staging setup from a finished project), oversized instances provisioned for a traffic spike that never repeated, storage nobody lifecycle-managed, and data-transfer patterns that charge you for architecture decisions made years ago.

The audit takes a day. Inventory every resource against its owner and purpose — anything with neither gets flagged. Check instance utilisation over 30 days; anything under 20% average is a right-sizing candidate. Apply lifecycle rules to storage (hot → cool → archive). And trace your transfer charges: cross-region chatter between services is often the single biggest fixable line.

Then make the savings structural, not a one-time cleanup: budget alerts at 80% of plan, tagging enforced in CI so every resource is born owned, and a monthly cost review with one owner. Autoscaling comes last — after you've watched real traffic patterns, not before.

We run this audit as a standard first step on managed engagements, and it typically recovers 20–40% on overspend accounts. The savings fund the monitoring and backup work that actually keeps the system safe — which is the correct order of operations.