Definition
Average time and behavioral change required for a merchant to be reclassified from a higher risk tier to a lower one, compared to the time and behavioral change required for escalation: measuring classification system asymmetry.
The Measurement Gap
The Measurement Gap
No metric tells you whether AI merchant risk tier classifications create self-reinforcing surveillance spirals. Once a merchant is flagged as high-risk by the behavioral scoring model, the system applies additional monitoring, lower hold thresholds, and more frequent review cycles. This increased scrutiny generates more data points, more anomaly detections, and more signals that confirm the high-risk classification: regardless of whether the merchant's actual behavior has improved.
The result is a one-way ratchet. Escalation is fast because any risk signal can trigger an upgrade to a higher tier. De-escalation is slow or nonexistent because the increased scrutiny generates a continuous stream of signals that justify maintaining the classification. A merchant who was correctly flagged six months ago for elevated chargebacks, and who has since reduced chargebacks to normal levels, may remain trapped in the high-risk tier because the additional monitoring still generates enough anomaly signals to prevent reclassification.
Merchant attrition rate captures the eventual outcome: merchants in high-risk tiers for extended periods eventually leave for a competing acquirer. But no metric captures the structural cause: that the classification system is asymmetric by design, with no mechanism ensuring that improvements in merchant behavior are recognized at the same speed as deteriorations. The gap is compounded by the fact that most risk platforms overwrite tier history rather than retaining it, so the asymmetry cannot even be analyzed retrospectively without deliberate instrumentation.
Decision Supported
Is the merchant risk classification system symmetric (improvements recognized as quickly as deteriorations), or has it created a one-way ratchet that traps reformed merchants in high-risk tiers?
Formula
Plain Language
Divide the median number of days it takes for a merchant to be de-escalated from a higher risk tier to a lower one by the median number of days it takes for escalation. A ratio of 1.0 means the system treats improvements and deteriorations symmetrically. A companion metric measures the behavioral change magnitude required in each direction.
Notation
RTEV = MedianDaysToDeescalate / MedianDaysToEscalate
Companion: BIR = BehavioralChange_deescalation / BehavioralChange_escalation
Variables
| Variable | Definition | Unit |
|---|---|---|
| MedianDaysToDeescalate | Median days from higher tier entry to lower tier reclassification | Days |
| MedianDaysToEscalate | Median days from last tier change to escalation trigger | Days |
| BehavioralChange_deescalation | Magnitude of chargeback/refund rate change triggering de-escalation | Percentage points |
| BehavioralChange_escalation | Magnitude of chargeback/refund rate change triggering escalation | Percentage points |
Proposed formula, pending validation pilots.