METHODOLOGY &
DATA SOURCES.

WHAT A CHAMELEON CARRIER IS

A chameleon carrier is a trucking company that accumulates safety violations, dissolves, and reincorporates under a new DOT number — same trucks, same drivers, same unsafe practices, clean safety record on paper. Using multiple DOT numbers to evade a negative compliance history is unlawful under federal law.

For the case for why this matters and how it shows up in public data, see Why CoopCheck Exists →. This page covers the underlying mechanics.

WHERE THE DATA COMES FROM

CoopCheck is built on the FMCSA CENSUS1 dataset — the federal government's official registry of all active and inactive motor carriers operating in the United States. This dataset is updated monthly by the Federal Motor Carrier Safety Administration and is public record.

Our current dataset covers 2.28 million carrier registrations, with FMCSA safety events through July 2026, refreshed monthly. Each record includes the carrier's legal name, DBA name, physical address, phone number, DOT number, fleet size, and operational status.

Safety history data — crashes, inspections, and violations — is sourced from the FMCSA SMS (Safety Measurement System) input files, also updated monthly. When a carrier is not yet in our snapshot database, CoopCheck falls back to a live FMCSA API lookup to surface real-time data.

Cluster analysis is performed against this dataset using shared contact information — phone numbers and physical addresses — as the primary signals of related carrier activity.

HOW WE DETECT CLUSTERS

CoopCheck identifies carrier clusters — groups of two or more registered carriers that share the same phone number or physical address. This shared-contact methodology is the same signal the FMCSA's own registration system uses to flag potential chameleon activity, but we surface it publicly and in real time.

A cluster does not prove fraud. Legitimate businesses can share addresses (office buildings, freight terminals) or phone numbers (brokerages, fleet management companies). What a cluster does is flag a pattern worthy of closer scrutiny — especially when combined with inactive DOT status, high carrier counts at a single address, or a history of safety violations.

Risk levels are assigned based on the number of carriers sharing a contact point:

SAFETY PERCENTILE SCORING

Each carrier receives a Chameleon Score — a letter grade (A through F) built on several core principles of the FMCSA's Safety Measurement System — a 24-month window, recency weighting, exposure normalization, peer comparison and percentile ranking: we rank every carrier against carriers that have a comparable safety record on its FMCSA behaviors, and the grade reflects its worst behavior. In plain language, the score reads "worse than X% of comparable carriers on [behavior]."

Power units means the carrier's current reported fleet size from its most recent FMCSA filing — not an average across the scoring window. That is a real limitation: a carrier that has shrunk since its crashes occurred divides them by a smaller fleet and scores worse than it should, and one that has grown scores better. FMCSA averages fleet size across the window instead. We have begun retaining monthly power-unit snapshots so we can do the same; until enough history accumulates, this page will keep saying which number we actually use.

Crashes FMCSA has determined were not preventable are excluded. Through its Crash Preventability Determination Program, FMCSA formally reviews crashes such as being struck while legally stopped, or struck by a driver crossing the median, and clears the carrier of them. Those crashes are excluded from the crash rate, from the fatality rate, and from the fatal-crash floor — 8.6% of all crashes on file carry such a determination. They remain visible on the carrier's record; they simply do not count against its grade. A crash with no determination is scored normally, which is not a finding of fault.

Five behavior categories are measured, each as a rate normalized for exposure. Events are recency-weighted — a crash or violation in the last 6 months counts triple, 6–12 months double, 12–24 months once, and anything older drops off — so the grade reflects a carrier's current risk, not stale history. A carrier is ranked on a behavior only once it has enough data for the comparison to be meaningful:

A carrier's score is the highest (worst) percentile among the behaviors it can be rated on. Grades are cutoffs on that percentile: F (99th+), D (94th+), C (84th+), B (55th+), otherwise A. Using per-unit and per-inspection rates — not raw totals — means a large fleet isn't penalized for its size, and a small carrier with a high crash rate isn't hidden by small numbers.

Two floors sit on top of the safety percentile. Any fatal crash raises a carrier to at least a B, so it never reads "Clean Record" — while a genuinely elevated fatal-crash rate is already reflected in the percentile above. A carrier in a confirmed reincarnation lineage is raised to at least a D. Confirmed means four things hold together: the carriers share a phone number and an exact street address, they trace to the same company officer, one of them had its federal carrier authority involuntarily revoked and shows no FMCSA inspections after that date, and another was registered within two years of it. A carrier that shares both a phone number and an exact street address with a small group of other carriers — between two and twenty — is raised to at least a B, never the D. Two independent identifiers are required, not one, and the group has to be small: a location used by dozens of registrations is a terminal or a mail drop, and the wider an identifier is shared the less it says about any single carrier. A shared phone number on its own does not change the grade: a number used by dozens or hundreds of carriers is usually a filing agent, dispatch service or insurance desk, and the more widely an identifier is shared the less it tells you about any single carrier. The cluster is still shown on the carrier's panel — we report the relationship without letting it stand in for a safety finding. The floor exists because chameleon carriers deliberately avoid accumulating violations before dissolving and re-registering, so a clean individual record is part of the pattern, not evidence of safety.

A behavior is only ranked once a carrier has enough of a record to rank it: 2 or more crashes for crash rate, 1 or more fatality for fatality rate, and 8 or more inspections for the out-of-service and violation rates. Below those thresholds there is no percentile to report, and a carrier that cannot be ranked is not thereby safe — it is unmeasured.

Carriers without a rankable record show one of three things, and the difference matters:

• NO HISTORY  No FMCSA crash or inspection history on record yet. Common for newly registered carriers. We know nothing about them.
• CLEAN · LIMITED  Zero crashes, zero out-of-service orders and zero violations across the inspections on file — but too few inspections to rank. Clean as far as it goes, and it does not go far.
• LIMITED  Crashes, out-of-service orders or violations are on record, but there are too few inspections to rank the carrier against peers.

All three are distinct from a low score. If a carrier is currently Out of Service despite a clean history, a regulatory warning is shown alongside the verdict so users have full context before making any decision.

That is how the grade is built. Whether it predicts anything →

DOES THE SCORE ACTUALLY PREDICT ANYTHING?

Everything above describes how the grade is built. None of it establishes that the grade is worth anything, and a methodology that only explains itself is not the same as one that has been checked. So we tested it, and this is what came back.

The method. We scored every carrier as it stood on 27 February 2026, using only FMCSA data published on or before that date — the archived crash, inspection and violation files from that month, and the fleet sizes recorded at that time. Nothing the model saw was newer than the scoring date. We then watched what happened to those carriers over the following 122 days and compared it against the grade each one had been given beforehand. The scores used for this test carry no floors: it measures the safety percentile on its own.

The result: the ranking holds. Among carriers that were actually operating, crashes over the following three months rose with every grade step, in order:

Out-of-service rate is the share of a carrier’s inspections that ended with a vehicle or driver placed out of service. Both outcome columns cover the 122 days after the scoring date.

An F carrier went on to crash about twice as often per truck as an A carrier. On crashes the order never breaks: every grade step is worse than the one before it.

The clearest number on this page is the last column. Over those same 122 days, more than half of every inspection an F carrier faced — 55% — ended with a truck or a driver ordered off the road. For a D carrier it was about two in five, for a C carrier under three in ten, and for an A or B carrier roughly one in six. That is an eight-fold spread from the top of the scale to the bottom, against under two-fold on crashes.

We publish both columns because they measure different things and one of them flatters us. Crashes are rare, so a three-month window catches few of them and the grades bunch together — D and F are barely apart on that column. Out-of-service orders are common enough to measure properly, and they are what most grades are actually built from. Reading only the crash column, as it is easy to do, understates what the grade separates.

A and B are close, and B is slightly better here. We are not going to pretend otherwise. The reason is fleet size: B in this cohort is dominated by large operators — 49 power units on average against 6 for A — and large fleets are inspected far more often and pass more often. The grade compares each carrier against peers with a comparable record rather than against the whole industry, so the top two grades sit close together and can swap depending on which outcome you measure. From C downward the ladder is steep and never breaks on either column.

How strong is that? Modest, and we would rather say so than round it up. Pick one carrier that crashed in those three months and one that did not, compare carriers of similar fleet size, and the score ranks the one that crashed as the worse carrier about 58 times out of 100. Fifty out of a hundred would be a coin toss. The grade is a real signal and it is not a prediction about any individual carrier — it tells you which way the odds lean, not what is going to happen.

It is strongest on the thing it mostly measures. Out-of-service history is the most predictive part of the model: a carrier's driver out-of-service ranking predicts its future driver out-of-service record at 66 times out of 100, the best of the five behaviors. Crash and violation rankings sit closer to 58. That ordering makes sense, because out-of-service and violation records are what most grades are built from — crashes are the deciding behavior for only about one carrier in fifteen at grade F.

One part does not hold up yet. The fatality rate ranking did not predict future fatalities in this test — it performed at chance. Fatalities are rare enough that 122 days is not long enough to tell whether that ranking works, so this is an open question rather than a negative result. We are not claiming the fatality percentile predicts anything until a longer window can show it does. The fatal-crash floor is a separate rule and is not affected: it raises a carrier that has already had a fatal crash, which is a matter of record rather than a prediction.

What this test cannot tell you. It covers one scoring date and one 122-day window, so it shows the ranking works; it does not show the ranking is stable across seasons or years. It measures the safety percentile only, so the reincarnation and shared-address floors are untested — establishing whether those predict anything needs registration histories rebuilt as they stood at the time, which we have not done. And 122 days is simply the most the published data allowed when we ran it: FMCSA releases records on a lag, and counting a partly reported final month would have flattered whichever carriers were active latest.

The window grows every month as more data is published. We will re-run this against the same scoring date as it lengthens, and update these numbers — including if they get worse.

First re-run, 17 August 2026. A monthly data refresh extended the observation window from 86 days to 122, a 42% increase, against the same carriers and the same scoring date. The ordering held and the figures barely moved: the discrimination measure went from 58 to 58 out of 100, out-of-service from 66 to 66, and an F carrier still crashed about twice as often per truck as an A. The fatality ranking moved from slightly below chance to exactly chance, which changes nothing about the claim we are not making for it. Numbers above are the longer window.

WHEN A CARRIER ISN'T IN OUR DATABASE

Our snapshot covers millions of carriers, but not every registered carrier will appear in search results. When a search returns no database matches, CoopCheck automatically queries the FMCSA live API to surface real-time carrier data.

Live results show a safety score, fleet size, and operational status computed on the fly from FMCSA's current data. A live result with no adverse events shows LIMITED LIVE DATA rather than a pass. The live API returns crashes, fatalities and out-of-service counts but not a violation count, so it can only tell you that none of those were found — it cannot rule out violations the way a database result can, and it should not be read as a clearance. CoopCheck also cross-references each live carrier's phone number against our cluster database in real time, so cluster tags appear on live results where a match is found. The safety score, however, is based on that carrier's own crash and inspection history — the full Chameleon Score, which ranks a carrier against its peers and applies the reincarnation-lineage floor, requires a complete database record.

Clicking any live result opens a full detail panel with address, fleet, operational status, cluster alert if applicable, and a direct link to the carrier's FMCSA safety record for deeper due diligence.

OPERATING STATUS BADGES

Live FMCSA lookups display an operating-status badge — a signal separate from the safety score, reporting what FMCSA's registration record says. It describes the state of the record, not a determination about whether a carrier may lawfully operate. That determination belongs to FMCSA and state enforcement, not to CoopCheck.

These badges come from FMCSA's live lookup at the moment you search, so they reflect the registration record as it stands then. For carriers in our own database, click Verify Auth ↗ in the detail panel to check current status at FMCSA directly — registration status can change between our monthly refreshes.

NO ACTIVE AUTHORITY

Separately, a carrier may show NO ACTIVE AUTHORITY in search results and an AUTHORITY block in its panel. That is a different signal, drawn from FMCSA's operating-authority registry and re-checked every morning rather than monthly. It means the registry lists no active for-hire authority for that carrier.

It is not an accusation. A carrier may operate lawfully without for-hire authority — hauling its own goods, or working within a single state. This is the status of the public record, not an enforcement determination. Where the panel notes that FMCSA inspections were recorded after authority became inactive, that is a separate factual observation about inspection dates; it is not a finding that the carrier was cited for operating without authority.

The absence of this flag is never itself a finding. Most carriers hold no for-hire authority at all and appear nowhere in that registry, so showing nothing means only that we have nothing to report.

CONNECTING LIVE RESULTS TO CLUSTERS

When a carrier appears via live FMCSA lookup rather than our database, CoopCheck now cross-references the carrier's phone number against our chameleon cluster database in real time. If a match is found, the carrier's result row shows a cluster tag — even though the carrier itself is not yet in our snapshot.

This means a live result can show LIMITED LIVE DATA + In Cluster simultaneously (a database result would read CLEAN · LIMITED or LIMITED instead). That combination is significant: it indicates a carrier with no individual safety history that shares contact information with a high-concentration cluster. This is precisely the chameleon pattern — a new or reincorporated entity at a flagged address with no safety history yet.

For brokers and insurers, this combination warrants additional due diligence even when the individual safety score is clean. The cluster context is the risk signal, not the score alone.

WHAT COOPCHECK IS NOT

INDUSTRY COVERAGE

The chameleon carrier problem has received growing attention from regulators, journalists, and the freight industry. Below are key resources for understanding the issue.