Measuring AI impact, closing the pilot-to-production value gap, and building the CFO conversation.
Turn AI initiatives into defensible investments — value hypotheses, NPV/IRR, TCO, attribution, and the CFO-grade conversation that gets funding released.
AI programme owners writing the business case, and finance business partners reviewing it.
The guides are numbered — read in order for the curriculum path, or jump straight to the one you need. Each card is self-contained.
Why the ROI conversation for AI breaks the old IT playbook, and the mental model that replaces it.
The six interlocking components of an AI business case that actually survives review.
Frameworks for sizing the value of an AI use case honestly, before you spend a dollar on it.
The full stack of AI costs, from tokens to change management, and where the biggest budget surprises hide.
A working TCO model for AI systems across build, run, and evolve phases over a realistic horizon.
The unit economics of inference and the levers (caching, routing, distillation) that reshape them.
Applying NPV, IRR, and payback to AI investments without pretending the future is deterministic.
Real-options thinking for AI investments when the future is genuinely unknowable, not just uncertain.
Which signals tell you AI is working before the annual metrics catch up, and how to instrument for them.
Why attribution is the quiet killer of AI business cases, and the counterfactual methods that make it defensible.
Measuring productivity impact from AI without the fantasy of 'hours saved' translating directly into money.
The archetypal ROI patterns for common AI use cases, and how to spot which one you are actually staring at.
Why agentic AI breaks the per-call ROI models and how to value multi-step economics honestly.
The AI vendor pricing landscape and how to negotiate contracts that survive a doubling of usage.
Chargeback and showback models for AI spend, and the governance patterns that make them stick.
The one-page structure for presenting AI business cases to CFOs and boards that actually gets funded.
The failure modes and anti-patterns that quietly kill AI business cases, and the signals of each.