Thank you, as a non-native English speaker this is extremely valuable feedback. I'll rework the text.
One idea is that could literally run a infrastructure diff of your architecture and your live environment, for example after a deployment or between two points in time: You may have resources left over by accident, or not captured in the design.