EC-301b · Module 1

What Executives Mean by ROI

4 min read

When an executive asks for ROI, they are not asking for a ratio. They are asking three questions: what do we get, when do we get it, and what could go wrong. A traditional ROI percentage answers none of those questions completely. A risk-adjusted value narrative answers all three.

The CFO definition of ROI in an AI context is: net benefit over the investment period, adjusted for the probability that the projected benefits materialize, divided by the total cost of ownership. That last part — total cost of ownership — is where most AI business cases fail. They include the software license and ignore the implementation cost, the training cost, the integration cost, the ongoing maintenance cost, and the opportunity cost of the people managing the system.

The executive version of the same concept is simpler: show me a number I believe, tell me when I'll see it, and tell me what happens if things go sideways. A business case that answers those three questions in the first two pages has done its job. Everything after that is supporting documentation.

Do This

  • State the net benefit in dollars with a defined time horizon: "$2.4M over 24 months"
  • Include total cost of ownership: license + implementation + training + integration + ongoing maintenance
  • State the risk-adjusted figure explicitly: "Projected at 75% confidence based on pilot data"
  • Lead with the payback period before the 3-year NPV — executives care when they see the money

Avoid This

  • Present ROI as a percentage without a dollar value: "247% ROI"
  • Include only the vendor license cost and call it the investment
  • Present a single best-case projection with no range or confidence statement
  • Bury the payback period on page 12 of the financial appendix