Cost per Request
The fully-loaded infrastructure cost of serving one request, and the unit that makes architectural cost decisions comparable.
Total cloud spend is not actionable — it goes up when the business grows, which is not a problem. Cost per request, or per tenant, or per transaction, is actionable: it says whether the architecture is getting more or less efficient independently of volume, and it is the number that tells you whether growth improves or worsens margin.
Computing it honestly means including everything: compute, storage, the database, egress, observability (frequently a surprise), third-party API calls, and support. A per-request cost built from compute alone is usually off by a factor of two or more.
Once it exists, architectural options become comparable in the same unit as product decisions, which is what lets an engineering trade-off be discussed with a CFO.