Unit Economics
Infrastructure cost expressed per unit of business value delivered, which reveals efficiency trends that absolute spend cannot.
Absolute cloud spend rising is not necessarily a problem. Spend per order, per active user, per gigabyte processed, per API call is the number that shows whether the system is becoming more or less efficient.
It changes the conversation. "Spend is up 40%" invites an indiscriminate cost-cutting exercise. "Cost per order fell 15% while volume rose 60%" is a demonstration of engineering value. "Cost per order rose 20%" is a specific problem with a findable cause.
Choosing the unit is the hard part, and it must be a business unit rather than a technical one. Cost per instance-hour is a technical metric nobody outside engineering can act on; cost per order is a number a product manager can weigh against margin.
Different parts of a business need different units: a marketplace uses cost per transaction; a SaaS product uses cost per tenant per month, which also exposes whether large customers are profitable; a data platform uses cost per terabyte processed.
The prerequisite is cost attribution by tagging, without which the metric cannot be computed below the level of the whole estate.
The strategic use: unit economics is what connects infrastructure decisions to gross margin, and it is the form in which engineering efficiency becomes legible to a finance function.