Run a serverless cost review with architecture context

A cost chart can show that spending changed. It cannot by itself tell a team whether the change is healthy growth, a retry loop, a retention mistake, or an intentional resilience trade-off.

Review the path, not only the service

Start with the workload path that produced the spend. Connect invocation volume, payload size, duration, concurrency, storage growth, and event fan-out to the customer action behind them. This often reveals that the expensive component is a downstream effect rather than the first line item that moved.

Then check the configuration choices around it: log retention, provisioned concurrency, retry settings, data transfer, and unused resources. Every recommendation should name the expected operational trade-off. Lowering retention may reduce cost while reducing investigation history; reducing retries may protect a dependency while increasing manual recovery.

The useful output is a prioritized decision, not a list of prices.