Insights
How to Calculate Cost Per Outcome for AI Agents
Cost per token is easy and insufficient. Finance cares about cost per outcome: the fully attributed AI labor cost to complete a defined business result — a resolved ticket, an underwriting memo, a reconciled invoice — versus the human baseline.
Define the outcome
Pick a unit leadership already recognizes. Ambiguous outcomes produce meaningless unit economics. Document inclusion rules: retries, tool calls, human handoffs, and failed runs.
Attribute the spend
Sum model, gateway, and tool costs for the workflow that produced the outcome. Attribute to the agent and owner. Without chargeback-ready attribution, your denominator is fiction.
Compare to the human baseline
Express AI cost per outcome next to fully loaded human cost for the same result. That is the CFO conversation — not a model leaderboard.
See also What Is Agent FinOps? and Measuring AI ROI Governance in Practice.
Not sure which of your AI costs are being booked? Run the free Agent Spend Assessment.
Onaro Meridian is FinOps for agentic AI: the system of record that attributes, controls and books what AI agents spend.
Brian Diamond
Brian Diamond is a fractional Chief AI Officer and founder of Onaro. He has spent 30 years running infrastructure operations and founded LANStatus, a Connecticut managed services provider and Microsoft partner, in 2001. He holds a Chief AI Officer certification and writes the CAIO Brief on AI leadership for finance and operations.