Cost reviews are one of the more common requests we get, usually after a surprising invoice. The temptation is to start cutting the biggest line items immediately. We don't — the biggest line item is often the thing least safe to touch without understanding why it's there.
Our order: first, find anything genuinely unused — orphaned volumes, idle load balancers, environments nobody remembers creating. That's free money with zero risk. Second, right-size what's oversized relative to its actual load, backed by real metrics, not guesses. Only after that do we look at architectural changes, which carry real risk and need to be scoped like any other project.
The unused-resource pass alone has paid for the entire engagement more than once. It's not exciting work, but it's the safest money on the table, and it should always come first.