Proof & ROI
The Most Dangerous AI Business Case Is the One That Cannot Tell the Difference Between Activity, Capacity and Cash
Why adoption, hours saved and user satisfaction are not the same thing as economic value.
Most AI business cases in Treasury are built on activity. Number of users. Prompts per week. Hours saved per analyst. These numbers are easy to collect and easy to present. They are also, on their own, almost useless for a CFO.
Activity tells you that something is being used. Capacity tells you that time has been released. Cash tells you that the P&L, the balance sheet or the risk position has actually changed. The three are connected — but not automatically.
Hours saved only become value when the organisation decides what to do with them: reduce external spend, absorb growth without hiring, or redeploy people to work that moves liquidity, funding cost or risk. If nobody makes that decision, the capacity quietly evaporates.
A useful discipline: before a pilot starts, write down which of the three levels the initiative is expected to move, how it will be measured, and who signs off that it did. If the answer is 'activity' only, call it what it is — an experiment, not a business case.
Continue the thinking
Personal views of the author; not the views of ING Bank or any other institution.