Data Methodology

1. Data sourcing

All data in Truck Graph originates exclusively from verified federal compliance records. We do not rely on carrier self-reporting, third-party aggregators, or unverified public submissions. Primary source datasets include:

• FMCSA Census (MC-150) — 1,985,646 carrier registrations covering legal names, DBA names, physical addresses, fleet size, driver counts, equipment types, and operating authority classifications.
• Authority history — complete grant, revocation, and reinstatement records per DOT number, spanning registrations from 1974 through 2026. Total: 634,633 revocation events.
• Enforcement actions — 63,000 out-of-service orders with effective dates, order types, and current enforcement status, updated continuously from FMCSA enforcement systems.
• Insurance filings — 900,000 insurance cancellation and lapse records drawn from BMC-91 and BMC-34 filings, covering 900,000 events across active and inactive carriers.
• Officer and principal records — registered officers, agents, and controlling principals tied to each DOT number, enabling cross-entity relationship mapping.
• Hazardous materials — HM permit registrations indicating carriers authorized to transport hazardous cargo under PHMSA and FMCSA joint authority.

All ingestion pipelines validate record integrity on arrival — deduplicating entries, normalizing address formats, and flagging anomalies before any record is promoted to the live index.

2. Data refresh and freshness

Truck Graph maintains a layered refresh schedule calibrated to the update cadence of each source dataset:

Every data block on Truck Graph displays a "last synced" timestamp. Carrier profile pages are server-rendered at request time against the live database — there is no stale cache layer between users and current federal records.

3. Composite risk score

Truck Graph's composite risk score (0–100, lower is safer) is computed from five independently weighted dimensions. The model is trained against verified crash outcomes and recalibrated quarterly to maintain predictive accuracy:

Scores require a minimum of 3 inspections within 24 months for statistical validity. Carriers below this threshold display "Insufficient data" rather than a potentially misleading score.

4. Network intelligence and relationship mapping

Individual carrier scores capture direct risk signals. But the most dangerous operators — particularly those attempting to evade enforcement by re-registering under new identities — require a different analytical lens: one that looks at connections between entities, not just the entities themselves.

Truck Graph's network intelligence layer models the entire carrier population as a relationship network, mapping connections across four shared-attribute dimensions:

• Shared officers and principals — carriers linked by common registered officers, controlling parties, or agents of record. A single officer appearing across multiple DOT numbers — especially where some are revoked — is a high-signal indicator.
• Shared physical address — carriers registered at identical physical addresses. High co-location density at a single address, particularly combined with a high revocation rate, signals a potential address of concern.
• Shared contact information — carriers sharing phone numbers or contact details across different legal entities and DOT registrations.
• Shared insurance relationships — carriers connected through common insurance providers or filing agents, which can reveal coordinated or fraudulent filing patterns.

These relationships are modeled with weighted connection strengths: a shared officer is weighted more heavily than a shared address, which in turn is weighted more heavily than a shared phone number. The network risk score for a given carrier is a function of both its direct signals and the weighted risk of its connected entities — carriers embedded in high-risk networks receive elevated network scores even if their own direct compliance record is clean.

The relationship graph currently indexes connections across the full 1,985,646-carrier dataset. Multi-hop traversal — following chains of connection up to 3 degrees — enables the system to surface hidden lineages where a chain of shared principals links a currently active carrier to enforcement actions several relationships removed. These extended paths are displayed in the Authority Pattern Alert section on each carrier profile.

5. Chameleon carrier detection

A chameleon carrier is one that surrenders or loses operating authority following an FMCSA enforcement action and then re-registers under a new legal name — retaining the same principals, address, equipment, or contact information — to obtain a clean compliance record. The FMCSA estimates this practice affects thousands of carriers annually and represents one of the highest-risk scenarios in freight engagement.

Truck Graph's detection algorithm evaluates five corroborating signals simultaneously:

• Address reuse after revocation — new authority registrations at a physical address previously associated with a revoked entity
• Officer overlap — shared registered principals between a currently active carrier and one or more revoked entities
• Temporal proximity — new authority grants within 12 months of a revocation involving shared attributes
• Fleet composition similarity — matching power unit counts and equipment types between the new entity and its apparent predecessor
• Contact information persistence — identical phone numbers or contact identifiers appearing across entities with different DOT numbers and names

A carrier is flagged when 3 or more signals match simultaneously with sufficient confidence. Of the 1,985,646 carriers currently indexed, 2,036 active carriers carry a prior revocation flag — meaning they were previously associated with a revoked authority before obtaining their current registration.

All flags are labeled "potential" indicators. Truck Graph does not assert legal violations — that determination belongs to FMCSA enforcement. Flagged carriers and their lineage appear in the chameleon carrier database.

6. Profile data tiers

Not all carriers in the federal database have the same depth of available information. Truck Graph assigns each carrier to a data quality tier that determines which analytical modules are available on its profile:

7. Current stage — v3.0

Truck Graph is currently operating at methodology version 3.0, representing a significant expansion from the original carrier profile and scoring system. The current stage of development includes:

8. Limitations and disclosures

• Risk scores are not safety certifications. A low composite score reflects favorable historical compliance indicators — it does not guarantee future performance or fitness for hire.
• Chameleon flags are algorithmic signals, not legal findings. A flagged carrier may be a legitimate new business that shares attributes with a revoked entity coincidentally. Always verify independently before freight decisions.
• Network scores propagate from connected entities. A carrier may receive an elevated network score because of who it is connected to — not because of its own direct compliance record. This is by design, but should be interpreted alongside the direct score.
• Data latency. Authority changes are reflected within 4 hours of FMCSA publication. Census updates (addresses, officer records, fleet size) may lag by up to 7 days.
• Equipment type inference. For carriers without explicit census data, equipment classification uses carrier operation type fields as a proxy. This is approximate and labeled accordingly on affected profiles.
• Inspection data gap. Carriers with fewer than 3 inspections in the trailing 24-month window do not receive a composite score. This disproportionately affects newer carriers — including some genuinely good operators with insufficient inspection history to score.

Questions about our methodology

For questions about data sourcing, scoring methodology, or to report a data discrepancy, contact our editorial team via truckgraph.com/contact. Methodology inquiries receive a response within 2 business days.