Why public safety data alone is not enough
BASIC scores and CSA percentiles miss fraud, double-brokering and reliability signals. Here is what they leave out - and how to fill the gaps.
Why Public Safety Data Alone Is Not Enough
CarrierTrust Insight Series | Data & Intelligence
Every freight broker in America has access to the same carrier safety data. FMCSA publishes it for free. The Safety Measurement System scores every carrier that crosses its data threshold. SAFER returns authority and insurance status in seconds. If public data were sufficient, carrier selection would be a solved problem, and the negligent-selection verdict docket would be thin.
It is not thin.
Public safety data is a necessary foundation, but it is not a sufficient one. Understanding what the data captures - and what it cannot - is essential to building a carrier qualification process that holds up in litigation, survives an FMCSA audit, and actually prevents bad loads from moving.
What BASIC Scores and CSA Percentiles Actually Measure
The Compliance, Safety, Accountability (CSA) program's seven Behavior Analysis and Safety Improvement Categories (BASICs) score carriers based on weighted violations and crashes drawn from roadside inspections and crash reports over a rolling 24-month window. Each carrier is then ranked percentile-by-percentile against similar carriers by operation type and size.
This is useful data. But its architecture creates structural blind spots that are relevant to every carrier qualification decision.
BASICs require sufficient inspections to score
A carrier must accumulate a minimum number of relevant inspections or crashes in a BASIC to generate a percentile score. Carriers with few or no recent inspections are "unrated" in those categories. According to FMCSA's own methodology documentation, a large share of active carriers - estimates based on MCMIS data suggest 50% or more - have insufficient data to generate a score in one or more BASICs.1
This creates an inverse visibility problem: the carriers most likely to be new, transient, or fraudulent - which by definition have minimal inspection history - are precisely the ones least likely to trigger BASIC-based alerts. The carriers public data surfaces most clearly are often not the highest-risk ones.
BASICs are lagging indicators
BASIC scores reflect what happened in the preceding 24 months at roadside inspections. They do not reflect a sudden deterioration in fleet maintenance after the last inspection, a change of ownership that brings new management and different safety culture, or a driver hired last month with a disqualifying record that has not yet been discovered in an inspection. A carrier's score on the day you check it may bear little relationship to how that carrier's equipment is actually maintained today.
CSA data has known accuracy limitations
The Government Accountability Office has published findings - in 2014 and again in 2019 - identifying data completeness and accuracy concerns that affect the reliability of BASIC scores as a comparative tool.2 State inspection reporting timelines vary, violation coding is inconsistent across jurisdictions, and crash causation determinations introduce noise that the percentile rankings cannot fully correct for. FMCSA has acknowledged these limitations publicly. They do not make BASIC scores useless, but they do mean scores should be treated as one signal among several, not a definitive safety verdict.
Formal safety ratings are issued to almost no one
The FMCSA's formal safety ratings - Satisfactory, Conditional, and Unsatisfactory - are issued only after a compliance review or investigation. The agency conducts compliance reviews for a fraction of active carriers annually. As a result, the overwhelming majority of active carriers - the percentage is consistently estimated above 95% - operate with an "unrated" status, having never received a formal safety determination.3 An unrated carrier is not a safe carrier; it is simply one the agency has not formally evaluated. Brokers who treat "unrated" as equivalent to "satisfactory" are operating under a significant misunderstanding of the regulatory framework.
What Public Data Completely Misses
Beyond the structural limitations of BASIC scores, there are entire risk categories that FMCSA public data does not address.
Fraud and identity theft
The fastest-growing carrier risk category is not measured by any public dataset. Strategic cargo theft through carrier identity fraud involves authority cloning (obtaining or fabricating a DOT and MC number closely resembling a legitimate carrier's), entity mimicry (registering a new entity under a name nearly identical to an established carrier and then obtaining legitimate authority), and compromised credentials (gaining access to a legitimate carrier's dispatch email or load board login).
None of these activities necessarily generate FMCSA violations. The fraudulent entity may have a clean authority record, a certificate of insurance that appears valid, and BASIC scores that are either pristine or simply absent. Public data shows a green light. The load disappears.
CargoNet (Verisk) documented a 57% year-over-year increase in cargo theft incidents in 2023, with fictitious pickup schemes - the carrier identity fraud category - accounting for the majority of high-value incidents.4 Losses are measured in hundreds of millions of dollars annually. None of this fraud was detectable through FMCSA public data alone.
Double-brokering
Double-brokering - where a carrier accepts a load and then re-brokers it to a third party without the originating broker's knowledge or authorization - creates compounding liability. The carrier actually performing the movement may not have been vetted by anyone. Insurance coverage for the re-brokered movement may be void. The originating broker retains legal responsibility for the load and the carrier that actually carries it.
FMCSA data contains no double-brokering signal. A carrier that routinely re-brokers will have its own clean BASIC scores and active insurance, with no public indication of its operational practice. Detection requires cross-referencing load board behavior, internal dispatch patterns, and industry intelligence - none of which is available through FMCSA's public systems.
Carrier performance and reliability
A carrier's SMS percentiles say nothing about whether it delivers on time, handles cargo claims responsibly, communicates proactively during service failures, or performs consistently across seasons and lanes. These are the signals that determine operational quality - and they are entirely absent from public safety data. They exist only in the internal records of brokers and shippers who have worked with the carrier, and in the carrier's own operational systems.
Insurance currency gaps
A certificate of insurance is a point-in-time document. It reflects policy status at the moment it was issued, not at the moment a broker checks it. FMCSA's insurance filing database is more reliable - it reflects what insurers have filed with the agency - but even it has reporting lag. A carrier can appear insured on a COI, appear insured on a partially current FMCSA filing, and be effectively uninsured at the moment of tender. This gap is invisible without a real-time query against live insurance filing status.
The Signal Gap Table
The table below maps each public-data blind spot to the supplemental signal category required to address it. No single commercial source provides all of these; a complete qualification process draws from multiple streams.
| Blind spot | Supplemental signal required |
|---|---|
| New or unrated carriers | Authority age, entity history, physical address verification |
| Identity fraud and cloning | Entity name + number cross-reference; fraud intelligence databases |
| Double-brokering | Load board pattern analysis; internal dispatch history; industry fraud registries |
| Insurance currency | Real-time FMCSA filing query at tender, not COI-only |
| Carrier performance | Connected operational data: TMS completion rates, on-time history |
| Chameleon carriers | Related-entity analysis; prior authority revocation history |
| Post-inspection deterioration | ELD compliance history; telematics signals; recency of inspections |
The Legal Standard Is "Reasonable" - and Reasonable Is Rising
The negligent-selection standard does not require brokers to be omniscient. Courts apply a reasonable care standard: did the broker take the steps that a reasonably prudent broker would take under the circumstances?
The practical problem is that what counts as "reasonable" is evolving rapidly. As the industry's understanding of fictitious pickup fraud, identity cloning, and real-time insurance verification has improved - and as courts, regulators, and shipper compliance programs have responded - the floor for reasonable care has risen. A qualification process that was defensible in 2019 may not be defensible in 2026.
The consequence for broker qualification programs is straightforward: the data stack needs to keep pace with the risk landscape, not just satisfy the regulatory minimum. Public data will always be part of that stack. The question is what sits alongside it - and whether those additional signal sources are integrated, documented, and consistently applied at the time of every tender.
Next in the CarrierTrust Insight Series: How Carriers Can Use Connected Data to Prove Trust - how ELD, GPS, TMS, and insurance integrations let carriers build and carry a verifiable trust credential across every broker relationship.
References
Footnotes
-
FMCSA Safety Measurement System Methodology, current edition. Available at: ai.fmcsa.dot.gov/SMS. The minimum data requirements section describes inspection thresholds by BASIC; carrier-level coverage rates are derived from FMCSA Motor Carrier Management Information System (MCMIS) population data. ↩
-
U.S. Government Accountability Office, Trucking Safety: Additional Research and Data Collection Could Help Identify Best Approaches for Oversight, GAO-19-585, September 2019. Available at gao.gov. See also GAO-14-114 (2014). ↩
-
FMCSA conducts compliance reviews for a small fraction of the approximately 600,000+ active carriers annually. The formal "unrated" percentage is consistently cited above 95% across industry sources drawing on MCMIS data. ↩
-
CargoNet (Verisk), 2023 Annual Cargo Theft Report, January 2024. Available at cargonet.com. ↩