The Metric to Anchor Your Agentic SOC Evaluation On

This analyst piece argues that agentic AI in SOCs has delivered real triage-speed gains but that product evaluations should prioritize detection-completeness—measured as median attacker dwell time—over simple per-ticket or alert-closure metrics; it recommends asking vendors for before-and-after dwell-time data, notes measurement and attribution challenges, and cites Mandiant and IBM industry data for context.