Personal productivity is not enterprise value
Eighty percent of people say AI has improved their own productivity; 37% of organizations can attribute any EBIT impact to it, flat year over year (McKinsey & Company, The state of AI in 2026: On the road to ROI, August 25, 2026, key takeaways and Exhibit 5). Global AI spending reaches $2.6 trillion in 2026, up 47% in one year (Gartner). Only 5% of integrated pilots are extracting meaningful value (MIT Project NANDA, 2025), and 38% of executives report no tangible financial return within twelve months (PwC, 2026). The spend is real; the attribution is not.
Self-report cannot produce it
Most programs measure activity: ideas submitted, pilots launched, licenses assigned. Value is asserted in a status field by the team that wants the funding. Blended ROI figures stack an opportunity estimate on a modeled one on a committed one and present the sum as an outcome — which is exactly why nobody believes the slide.
Five value states, never summed
Every initiative carries its value in five separate states — opportunity, modeled, committed, realized, verified — each figure with source, method, owner and as-of date. Adoption is read from the systems of work, not surveyed. The record stays open until a named person in Finance puts their name on the verified number, and only verified numbers reach the board.
What we will ask you for
- Your current portfolio list — however informal.
- The figure you last showed the board, and how it was built.
- The three initiatives you would most like to be able to prove.
The reports behind the question
- McKinsey · The state of AI in 2026: On the road to ROI (Aug 25, 2026)
- BCG · AI Talk Is Cheap. Value Creation Is Rare.
- Accenture · Pulse of Change
- Gartner · AI spending forecast 2026
- MIT Project NANDA · 2025
- PwC · GenAI Leaders Survey 2026