Economics
Investment cases that can be inspected, not just presented
Status disclosure
Planned study
Public research brief. No findings are published as completed fieldwork. The study plan may still change. Fieldwork has not begun.
Which assumptions, if wrong, would stop the spend before a broader rollout?
Consequential AI and operations bets often arrive with a single projected return and no way to see what would falsify it. This brief treats the investment case as an inspectable object: assumptions, ranges, denominators, and a stop condition that leadership can use without waiting for a post-hoc story.
The business case is a deck, not a model anyone can reopen after the meeting.
Value is claimed from time saved or error reduced without a baseline the operators recognize.
Sensitivity lives in an appendix, so the decision is made on the happy-path number.
How we would work it
Put assumptions, ranges, and kill criteria beside any projected value.
Separate observed operating data from management targets and vendor claims.
Fund the next increment only when the metric that would change is already being measured.
What has to be true
- Can a skeptical operator trace the number back to a source, a denominator, and an owner?
- Which single wrong assumption would reverse the recommendation?
- Is the next dollar buying evidence, or buying a larger version of the same untested case?