Why isn’t our AI investment paying off?
Usually because AI was the starting point instead of the outcome. The fix starts with what the investment was supposed to produce, and whether anyone measured it.
Which of these happened?
- The promised productivity gains never showed up
- The pilot worked, but production didn’t
- AI spend is growing faster than the benefit
- The board wants an ROI number no one can produce
- Nobody defined what success looks like
Why it happens
I don’t just look for where AI can go. I look for where it belongs.
Most AI that doesn’t pay back isn’t a technology problem. The work wasn’t ready for it, the economics were never modeled, or adoption never happened. The research says that is the norm, not the exception.
Sources: MIT Project NANDA, The GenAI Divide: State of AI in Business 2025 (preliminary); S&P Global Market Intelligence (2025).
The organizations that do see a return start with a business outcome, fit AI into how the work actually runs, and measure it from day one.
What I do
Model what the investment was supposed to produce, what it is producing, and the gap between them. Then find the cause and decide whether to adjust it, redirect it, or stop it.
FAQ
Is stopping the investment ever the right answer?
Yes. Redirecting or stopping an AI investment is a legitimate outcome of the diagnostic.
What does good AI ROI look like?
Measurable gains in output, cost, cycle time or error rate that exceed the full cost of the AI, including running it.
How long before we see a return?
It depends on what it was built to do and whether it was placed correctly. The diagnostic gives you a realistic answer.