Every team running AI agents needs one number to answer: is our AI spend worth it? We propose the Agentic Cost Ratio (ACR).
The formula
ACR = Value Produced / Cost Incurred
Value depends on the agent's role: tickets resolved × cost per human ticket, pipeline generated, hours of analysis saved. Cost is everything: LLM calls + tools + compute + overhead.
What good looks like
ACR above 3.0: Agent produces 3x more value than it costs. Healthy. ACR 1.0-3.0: Break-even to moderate ROI. Worth optimizing. ACR below 1.0: Agent costs more than value created. Time to fix or kill.
Why existing metrics fall short
Cost per token is too granular — doesn't tell you if tokens produced value. Total AI spend is too broad — doesn't tell you which agents are efficient. Monthly bills are backward-looking — tell you what happened, not what to do.
ACR bridges cost and value. It lets an EM tell the CFO: "our support agent has ACR 4.2 — every dollar returns $4.20 in value."
How AgentCostPilot helps
Value Reports calculate ACR automatically. Connect cost data (we have it), input value metrics (tasks completed, hours saved), and the report generates per-agent ACR with trends. Export as PDF for stakeholder reviews.