Unit Economics and KPIs

Unit economics divides cost by a business driver — cost per order, per active user, per GB processed — so spend can be judged against value; FinOps KPIs such as allocation coverage, commitment coverage and forecast variance track the practice itself.

"Cloud cost went up 20%" says nothing on its own. "Cost per order went down from $0.042 to $0.036 while orders doubled" says the business is getting more efficient. That is unit economics.

flowchart LR
  COST["Allocated cost of the product"] --> U["Unit cost = cost ÷ units"]
  DRV["Business driver: orders, users, GB"] --> U
  U --> TR["Trend over time"]
  TR --> DEC["Decisions: pricing, architecture, priorities"]
  REV["Revenue per unit"] --> M["Margin per unit"]
  U --> M
  %% step COST: Start with the fully allocated cost of a product, including its share of shared costs.
  %% step DRV: Choose a driver that grows with value and that the business already tracks.
  %% step U: Unit cost is simply the cost divided by the number of units in the same period.
  %% step TR: The trend matters more than one value: is each unit getting cheaper as we grow?
  %% step REV: Combine with revenue per unit to see the margin each unit contributes.
  %% step DEC: Product and engineering use it to price plans and to rank optimisation work.

Choosing a unit

Business	Possible unit	Why
E-commerce	Cost per order	Grows with revenue
SaaS	Cost per active customer or tenant	Ties to subscription revenue
Media / data	Cost per stream hour or per TB processed	Ties to delivered volume
Internal platform	Cost per deployment or per developer	Shows platform efficiency

A good unit is understood by the business, measured reliably, and moves with cost. Start with one, not ten.

FinOps KPIs (measuring the practice)

KPI	Formula / meaning
Allocation coverage	Allocated cost ÷ total cost (target: close to 100%)
Commitment coverage	Eligible usage covered by Savings Plans/RIs ÷ eligible usage
Commitment utilisation	Commitment used ÷ commitment purchased
Forecast variance	(Actual − forecast) ÷ forecast
Savings realised	Savings implemented ÷ savings identified
Unit cost trend	Change in cost per unit over time

Benchmarking compares these across teams or against peers to show where to focus.

Worked example

October: cost of the checkout product $12,600, orders 350,000 → $0.036 per order. September: $10,500 and 250,000 orders → $0.042. Total cost rose 20%, but unit cost fell 14%: a success story to tell, not a problem.

Traps: choosing a unit nobody outside engineering understands; computing unit cost from unallocated or partial cost; and celebrating lower unit cost while total cost exceeds what the business can afford.