The Metrics Brothers (fka SaaS Talk) A New Agentic AI Metric: Containment Rate
10 snips
Nov 20, 2025 Explore the emerging metric of Containment Rate, which measures how many tasks an AI agent completes independently. Discover how it differs from the traditional deflection rate and why precise definitions of 'resolved' are crucial. The hosts dive into the importance of connecting this metric to ROI, highlighting both benefits and investments. They also share benchmarks across various sectors and discuss the future of AI agents alongside customer portals. With humor and insight, they offer valuable recommendations for maximizing AI efficiency.
AI Snips
Chapters
Transcript
Episode notes
Containment Rate Defined
- Containment rate measures the percent of tasks an AI agent completes end-to-end without human handoff.
- It generalizes deflection rate from IVR/chatbots to agentic AI handling autonomous workflows.
Containment Versus Deflection
- Deflection can mask poor user experience when users give up before resolution.
- Containment focuses on true completion rather than merely preventing human involvement.
Apply Metric To Task-Based Agents
- Restrict containment rate to tasks where agents replace humans and escalations are possible.
- Apply it to customer support, SDRs, IT, HR, and back-office workflows with clear escalation paths.