The recurring ones that never appear on a quote: data egress and inter-region transfer, idle non-production environments, unused commitment coverage, observability and logging growth, and the engineering hours spent on manual cost governance. Each is small monthly and large annually.

Budgets are built from list prices for planned workloads. The missed costs are usage-driven — they scale with traffic, team growth, and change velocity rather than with the original plan, so they only become visible once the system is live and evolving.

Track leading indicators from month one: egress as a share of the bill, non-production spend outside working hours, commitment utilisation, and log-storage growth. The key practice is giving every recurring cost an owner and a review date, so no line item drifts unexamined.

Native tools (AWS Cost Explorer, Azure Cost Management, GCP Billing) show where spend goes. Governance platforms like Cloudeva.ai add the missing layer: connecting each cost to the decision that created it, so finance sees not just the number but who owns it and when it will be reviewed.