Definition
CSARF measures the change in security analyst investigation depth and response time as a function of cumulative false positive exposure, quantifying whether the human security layer is degrading due to alert fatigue.
The Measurement Gap
Measurement Gap and Decision Supported
The Invisible Failure Mode: A water utility's OT security analyst receives an average of 47 alerts per week from the SCADA anomaly detection platform. Over 18 months, the analyst's average investigation time per alert has decreased from 22 minutes to 9 minutes, and their response time to high-severity alerts has increased from 8 minutes to 34 minutes. The analyst is not less competent: they are rationally adapting to an alert volume that includes 73% false positives. They have learned which alert patterns are almost always false, and they apply that pattern recognition to triage faster. The problem is that the one genuine cyber event that eventually arrives will match a pattern the analyst has learned to dismiss quickly. The analyst's degraded response is invisible in security metrics because no real incidents have occurred to reveal it.
What Existing Metrics Miss:
Security operations metrics track alert volume, resolution time, and escalation rates but not the behavioral degradation in investigation quality. An analyst who closes alerts faster might appear more efficient when they are actually less thorough. Without CSARF, the progressive degradation of the human security layer is invisible until a missed incident reveals it.
The Decision CSARF Enables:
Is the OT security platform's alert volume degrading the human response layer to the point where analyst fatigue creates a security gap, and should the platform's alert thresholds be tightened or alert suppression rules added to reduce the false positive burden?
Cross-References:
CSARF connects directly to Alert Signal-to-Noise Ratio (ASNR, GMI-WT-010) from the maintenance KPI set and to Customer Leak Alert Positive Predictive Value (CLAPV, GMI-WT-024). All three address the same fundamental problem (false positive volume degrading the response quality of the human in the loop) in different operational domains. ASNR measures internal operational alert quality, CLAPV measures customer-facing alert quality, and CSARF measures security alert impact on analyst behavior. CSARF also connects upstream to SAAA (GMI-WT-031): high false positive rates that drive SAAA's operational misattribution category are the same false positives driving CSARF degradation.
Formula
Plain language: Track two metrics over time for each analyst: average investigation time per alert (declining = less thorough investigation) and time-to-first-action on high-severity alerts (increasing = slower response). Correlate both with cumulative false positive exposure.
Notation:
CSARF = (Investigation_depth_change x 0.5) + (Response_time_change x 0.5)
Where:
- Investigation_depth_change = (Avg_investigation_time_current - Avg_investigation_time_baseline) / Avg_investigation_time_baseline (negative = degrading)
- Response_time_change = (Avg_response_time_high_severity_current - Avg_response_time_high_severity_baseline) / Avg_response_time_high_severity_baseline (positive = degrading)
- baseline = first 3 months of deployment or pre-AI manual investigation times
Units: Composite index. Negative values = degrading (investigation getting shallower AND/OR response getting slower). Zero = stable. Positive = improving (unlikely without intervention).
Proposed formula, pending validation pilots.