Back to hub
REC-30215Generated by margin-optimizer v2.4 · 4m ago
Auto-approve accessorial dispute batch
Batch B-2298
Margin uplift
+0.4pp
+$14.1K
Service impact
0.0pp
OTD-equivalent
Confidence
95%
decision threshold 60%
Risk-adjusted score
91
0–100 composite
ROI multiple
22.0x
vs implementation cost
Expiry
2h 41m
after which trajectory shifts
AI Rationale
why this recommendation162 invoices match prior-won dispute pattern with precision 0.97 across last 6 months. Auto-approval keeps within human-in-the-loop policy (batch < 200 invoices).
SHAP contributors
Pattern match precision 0.97
+0.30
Win rate history
+0.15
Trajectory Impact
if accepted vs baseline14d projected
+0.4pp
Risk-adj
91/100
Expires
2h 41m
Counterfactual Comparison
alternative optionsOption
Margin
Service
Conf
Auto-approve batch
High precision, within policy cap.
+0.4pp
0.0pp
95%
Manual review
Same outcome; delays recovery 3-5d.
+0.4pp
0.0pp
99%
Constraint Compliance
policy & contractBatch ≤ 200 invoices
162Precision ≥ 0.90
0.97Acceptance Workflow
human-in-the-loop + RLHF capture RLHF signal capture
Audit & Provenance
data → model → recommendationModel
margin-optimizer v2.4
Training cutoff
2025-05-12
Features used
48
Causal uplift method
DoubleML
Counterfactual sample
2,418 sims
Validator
constraint-suite v1.7
Linked alerts
ALR-9912, ALR-9898
Approver pool
Logistics Director, Finance