Definition
Change in shipper volume commitment or tender frequency following AI routing decisions that prioritized rail revenue over door-to-door service quality, measuring whether modal optimization retains or repels customers.
The Measurement Gap
The Measurement Gap
Traditional intermodal commercial metrics track volume, revenue per unit, and customer churn at the aggregate level. None of them connect specific AI routing decisions (the ones that chose rail over truck for a particular lane) to subsequent customer behavior on that lane. The routing optimizer maximizes current-shipment revenue, but nobody measures whether those decisions are systematically driving customers back to trucking.
This gap matters because intermodal routing optimizers make thousands of modal-selection decisions daily, and each one is a small experiment in customer tolerance. A customer whose shipment gets routed via rail on a lane where they previously used truck is being "converted." If the rail experience meets expectations, they stay. If transit time, reliability, or damage rates disappoint, they revert, and they take future volume with them.
Decision Supported:
Are routing decisions that maximize current-quarter rail revenue eroding the customer relationships that generate future-quarter revenue? ICRS answers this by tracking converted customers at the lane level and measuring whether they stick with rail or flee back to truck.
Formula
Plain Language: Identify shipments where the AI router selected rail over truck for a lane where the customer had previously used truck. Track the customer's subsequent behavior on that lane: did they continue using rail, revert to truck, or reduce total volume?
Notation:
ICRS = Retained_customers / Total_converted_customers
Variables:
| Variable | Definition | Unit | Source | Classification |
|---|---|---|---|---|
| Retained_customers | Customers who continued rail usage on the converted lane in the follow-on period | Count | Commercial system | DERIVED |
| Total_converted_customers | Customers whose shipments were routed via rail by the AI when they had previously used truck on that lane | Count | Commercial system + routing optimizer | DERIVED |
Sign Convention: 1.0 = all converted customers stayed (ideal). Lower values indicate customer churn attributable to routing decisions. Track separately by reason for churn: transit time, reliability, damage, cost.
Edge Cases:
- Customers with contractual volume commitments may continue using rail even if dissatisfied, until the contract expires. Measure ICRS at contract renewal, not just quarter-to-quarter.
- Multiple lanes per customer: a customer may increase rail usage on one lane while abandoning another. Measure at the lane level, not the customer level, then aggregate.
- Minimum sample: 50 converted customer-lanes per measurement period.
Proposed formula, pending validation pilots.