Nothing here is verified yet. Content is migrated or drafted and has not been reviewed by a practitioner. Assertions marked like this are unchecked.
FinAtlas
Taskunverified

Unit economics / cost per unit

Express the business as economics of one unit (customer, transaction, seat, ride) to test whether scale improves or degrades profitability.

varies by archetype

3

The task exists in this business archetype but behaves differently. The note is the point of the edge: it must say how it differs, not merely that it does.

  • AI / compute infrastructureArchetypeunverified

    Cost per token or per request depends on accelerator depreciation life, cluster utilisation and power price — three assumptions set by finance rather than observed. A change of one year in assumed useful life can move gross margin by tens of points, so the unit economic is only as credible as the depreciation policy behind it.

  • Unit cost is dominated by fab utilisation and yield rather than by any variable input. At low utilisation, fixed depreciation per wafer swamps every other line, so cost per unit is really a capacity question and cost-reduction programmes aimed at materials move almost nothing.

  • Services / consultingArchetypeunverified

    The unit is a billable hour, so unit economics collapse into utilisation, bill rate and realisation. There is no scale curve: serving twice the revenue requires close to twice the people unless leverage or price changes.

produces

2

The modeling object or downloadable deliverable the task creates.

consumes

2

A canonical data object the task needs as input.

owned by

1

The business-partner org unit accountable for the task.

runs in

1

The end-to-end shared-services process the task sits inside.

serves

1

The verb-level intent the task exists to satisfy.