After windshield replacement or front-end body work, shops often return a vehicle with a clean scan report and a green OEM health dashboard. That is not the same as a calibrated ADAS stack under real driving residuals. Fleets need a shadow scoring layer that compares telematics you already export against repair-event context — without ECU writes — before vehicles return to revenue service.
The gap between scan pass and road truth
Static target routines prove the scan tool communicated with ECUs. They do not continuously prove camera-to-vehicle extrinsics, radar boresight, or fusion consistency across temperature, load, and vibration cycles. A vehicle can pass in bay lighting and drift within the first hundred revenue miles.
Fleet safety teams see this as liability exposure: the repair record says calibrated, the telematics story says nominal, and the first near-miss event has no signed evidence chain tying repair timing to residual elevation.
Why dashboards lie quietly
OEM and telematics dashboards aggregate health at coarse intervals. Missing calibration events, partial procedures, or skipped dynamic steps often surface only as subtle residual elevation — not as a hard fault code. Without cross-modal scoring, a forward camera can be several tenths of a degree off while lane-keeping and AEB still appear functional until edge cases stack.
That is the core detection problem NADIR pilots target: find miscalibration after repair before it becomes a dispatch incident or insurer dispute.
Shadow scoring after repair events
Shadow mode ingests batch telematics (CSV, JSONL, or API), joins repair order timestamps, and scores Pulse-tier residuals per VIN. CRITICAL rows enter a Console queue; signed evidence bundles capture model version, source signals, and tier transitions for audit — with no automated dispatch holds until stakeholders trust alert rates.
The operational contract is simple: every post-repair VIN gets a detection window; MTTFF-RE measures how fast CRITICAL flags appear after the RO close timestamp.
What fleets should require from partners
- Repair-event feed or weekly RO CSV with VIN and close time
- Telematics export cadence that covers the first seven days back on route
- Shadow-only pilot language in LOI — advisory analytics, not ASIL, no ECU writes
- Week-4 readout with MTTFF-RE distribution, not just tier histograms
Next steps
Start with the windshield recalibration guide, watch the product walkthrough, then scope a 4-week shadow pilot on 50–150 ADAS-equipped vehicles.