McKinsey's research into AI measurement practice identifies a consistent differentiator between organisations capturing AI value and those that are not: top-performing organisations measure across three time horizons. Only 18% of companies do this. The majority measure only the immediate horizon, time saved, tasks completed, adoption rates, and miss the near-term and strategic value that the same investment is generating. The immediate metrics are real. They are not the full picture. Organisations that measure only immediately are systematically undervaluing what they have built and making subsequent investment decisions on an incomplete evidence base.
Three measurement horizons capture the full value of an AI programme. The immediate horizon measures operational impact: time saved, error rates reduced, output volumes increased. These are visible within weeks and useful for demonstrating early traction. The near-term horizon measures capability building: workflows documented and transferable, governance frameworks operational, measurement baselines established. These are visible within months and indicate whether the AI capability is becoming institutional. The strategic horizon measures compounding: whether AI is generating competitive positions, customer outcomes or organisational capabilities that were not previously achievable. These are visible at programme close and beyond.
A programme assessed only on the immediate horizon will be undervalued at the point when near-term and strategic value is starting to accumulate. The board sees time saved and thinks the return is limited. The operating model it has built allows the organisation to do something strategically new. Both are true. The complete measurement picture shows both.
The organisations that can demonstrate the full return of AI at programme close are the ones that established measurement across all three horizons before deployment began. The 18% that do this establish one metric for the immediate horizon: operational impact measurable within weeks. One for the near-term horizon: capability building measurable within months. And one for the strategic horizon: a business outcome that the programme is designed to make possible, measurable at programme close and beyond. This three-horizon measurement structure is not decoration. It determines what return the board sees at close. Single-horizon measurement systems reveal only immediate productivity gains and systematically obscure the institutional capability that is accumulating in the other two horizons. Organisations that can show the full picture consistently generate more investment for subsequent AI programmes because they have demonstrated value beyond the adoption numbers that characterise the majority of AI reports.