ROI
Return on investment: the value gained relative to the cost. For AI it is easy to overstate by counting local savings and ignoring supervision, exceptions and recovery.
ROI compares what an investment returns with what it costs. In AI programs it is usually presented as time saved, tickets deflected or calls automated against the spend on models and tools.
The usual weakness is the denominator and the boundaries. A model's cost is visible; the human attention needed to make results dependable is not. Once supervision load is visible, AI ROI becomes a little harder to fake. Similarly, costs can move between budgets, so a saving in one place may be paid for in another: use the cost-migration test.
Better foundations
- Count the whole system: total cost of ownership.
- Measure per result: cost per accepted outcome.
- Remember the future: in customer service, the ROI model needs a memory of what today's interaction does to tomorrow's channel choice (repeat-channel economics).
- Do not equate usage with value: high AI adoption can coexist with low returns.
Read more in The hidden metric in AI automation is supervision.
Related terms
Total cost of ownership
The full cost of running an AI system across its life: model and inference spend plus integration, human review, exceptions, recovery, governance and maintenance — not just the price per call.
Cost per accepted outcome
The total cost of getting one AI-assisted result that is actually accepted and used, counting model spend, retries, human review, exceptions and recovery — a truer measure than cost per call.
Supervision load
The human attention an AI system still consumes — context supply, review, approval, correction and exception handling — measured per dependable, completed unit of work.
AI adoption
How widely an organization's people use AI tools — licences, active users, use cases. It is rising faster than enterprise value, because adoption is not the same as changing how the work gets done.
Used in these essays
The biggest AI mistake companies make is starting with AI
An AI-first mandate can turn discovery into justification. The missing discipline is preserving the right to choose a better, simpler answer.
Your AI program can be efficient and still be strategically irrelevant
AI can make low-value work dramatically cheaper. That is still low-value work. The portfolio question is not only what became faster, but whether the work deserved investment in the first place.
Your AI agent estate is becoming a capital allocation problem
Once agents multiply across workflows, the management problem changes: which agents deserve more capacity, which should be merged or constrained, and which should be retired?
Stop adopting AI
AI adoption is rising faster than enterprise value because companies keep installing new intelligence inside old operating models.
A cheaper AI model can move the cost instead of removing it
A lower model bill can hide a higher workflow bill. The useful AI TCO question is not only what got cheaper, but where the cost moved.
The hidden metric in AI automation is supervision
As AI moves from assisting work to leading it, hours saved stop telling the whole story. The scarce resource shifts to human supervision: approvals, exceptions, context and judgment.
Your chatbot is borrowing from the next interaction
AI customer service is usually measured one interaction at a time. But a failed automated interaction can change which channel a customer chooses next time. That makes future adoption part of the economics, not a separate trust metric.