Carrier Vetting After Montgomery (Parts 3-4)

For Day one of this series see 7/23/2026 post

Part III. What Good Looks Like

9. Twenty minutes in the yard

Everything in the federal record is downstream of the yard. The inspection data, the crash history, the out-of-service rate, all of it’s the paper shadow of physical decisions somebody made about trucks, drivers, and money, months before enforcement ever saw the result. When I walk a yard for an underwriter, I’m not gathering different information than the roadside record holds. I’m gathering it earlier, before it becomes a violation, and I’m gathering the parts the roadside record structurally can’t capture. Twenty minutes on the ground tells you things that two hours in the data won’t, and the order you look at them in matters.

Start before you get out of the truck. The gate and the fence line tell you whether anybody thinks the equipment is worth protecting. A yard full of loaded trailers behind a chain that anybody with bolt cutters owns is a cargo theft claim that hasn’t happened yet, and it’s also a statement about how the company thinks about risk generally, because physical security is the cheapest risk control there is and they didn’t buy it.

Walk the dead line first, not the active equipment. Every yard has one: the row of trailers and tractors nobody has moved in months, and the dead line is the maintenance program’s confession. Look at what’s sitting there and why. Blown moon roofs, flat-spotted tires, trailers with the doors roped shut. A carrier that cannibalizes its dead line for parts is a carrier managing cash, not maintenance, and you’ll find the results of that management in the vehicle out-of-service rate about six months from now. The deadline also tells you fleet age honestly, which the MCS-150 doesn’t.

Then look at the tires on the live equipment, because tires are money and money is the truth. Matched tires at legal tread on the drive axles cost real dollars and mean somebody is spending on the thing that touches the road. Recaps on steer axles, mixed brands worn to different depths on the same tandem, weather cracking on trailer tires: each one is a maintenance interval that got skipped, and each one is visible from ten feet away without a gauge. You can read a maintenance budget off a tandem faster than off a financial statement.

The shop, if there’s a shop, answers the program question. A real preventive maintenance operation has work orders you can see, a parts room with an inventory system, fluids stored like somebody expects an EPA visit, and a torque wrench that’s been calibrated within the last year. Ask when the last PM was done on any unit you point at, and watch whether the answer comes from a system or from memory. A carrier that outsources all maintenance isn’t disqualified; plenty of good small fleets run that way, but then the question moves to the vendor invoices and whether repair closure is verified or assumed.

Watch the people, last and longest. Who walks up to you and how fast tells you whether anybody owns the yard. Drivers doing real pre-trips versus drivers doing walk-arounds with a coffee tell you whether the inspection culture is practice or paperwork. The dispatch office wall tells you what gets measured, because companies post what they care about, and a wall covered in revenue-per-truck with no safety scoreboard anywhere tells you its priorities are in laminate.

None of this is available to a broker covering a load at 4:45 on a Friday, and that’s fine. The yard walk is Tier 3 work, the underwriting depth, the thing you do for core carriers, dedicated freight, and any account you’re about to write. The reason it belongs in this guide anyway is that everything the data shows you is a proxy for what the yard would show you, and once you’ve walked enough yards, you read the data differently. A high vehicle out-of-service rate stops being a statistic and becomes the dead line you can picture. The carrier profile becomes a photograph of a place, and you get much harder to fool.

10. Driver qualification that’s real versus a file cabinet

The regulation is 49 CFR Part 391, and the industry has spent fifty years learning to satisfy it without performing it. A driver qualification file is a folder with roughly a dozen required items: the application, the motor vehicle record pulled at hire and annually after, the road test or its equivalent, the medical certificate, the previous employer inquiries, and the annual review of the driving record. Every audit checks that the folder exists and the items are in it. Almost nothing checks whether any of it was an actual decision.

The difference between a real DQ process and a file cabinet is whether anything in the file could ever cause a no. Pull ten files at a carrier and read them as a set. If every application shows the driver was hired within a day or two of applying, the previous employer inquiries came back after the hire date, and no file in the building documents a rejected applicant, the process is a formality that has never stopped anyone. A real process leaves evidence of friction: an applicant file marked declined, an MVR that triggered a conditional hire with a documented restriction, a road test with actual scored deficiencies instead of a checked box.

The annual MVR review is where the paper and the road part company most reliably, and it’s the contradiction I look for first because it’s the one the roadside record can confirm or destroy. A carrier that certifies it reviews every driver’s record annually, while its inspection history shows a steady stream of driver violations from the same CDL numbers, has a review process that reads records and does nothing about them. That gap, between the attested control and the observed outcome, is the single most useful credibility test in the entire assessment, because it doesn’t require you to judge the program. The program judged itself.

The drug and alcohol program has the same split. The regulation requires pre-employment testing, random testing at prescribed rates, and queries against the FMCSA Drug and Alcohol Clearinghouse at hire and annually. A real program produces a paper trail of random selections generated by a third party on a schedule the carrier doesn’t control. A cosmetic one produces round numbers, tests clustered in the same week each quarter, and a Clearinghouse query log with gaps. The Clearinghouse matters more than the industry has absorbed: since 2020 it’s the national record of violations and return-to-duty status, and a carrier that isn’t querying it annually is certifying drivers it hasn’t checked against the one database built to catch the drivers who move.

Turnover is the number that explains everything else, and it isn’t in any federal record. Truckload turnover at large carriers has run near or above 90% annually for most of two decades, and at the bottom of the market it exceeds 100%, which means the average seat changes occupants more than once a year. A carrier at that churn rate has no accumulated knowledge of its drivers, is perpetually hiring, and is under perpetual pressure to lower the bar to fill trucks. Ask the turnover number directly. A carrier that knows it, tracks it, and can tell you what it was last year and what they changed is managing the thing. A carrier that doesn’t know it is being managed by it.

11. Maintenance as a program

The regulation is Part 396 and it asks for systematic inspection, repair, and maintenance. The word doing the work in that sentence is systematic, and it’s the word that separates the two kinds of carriers you’ll meet.

A reactive operation fixes trucks when they break or when a roadside inspection makes them. Its maintenance record is a stack of repair invoices with no pattern, its vehicle out-of-service rate runs above the national average, which has sat in the low twenty percent range for years, and its violations skew toward the things that fail gradually and visibly: brakes out of adjustment, tires, lights, air leaks. Brake and tire violations are the tell because nothing about them is sudden. A brake goes out of adjustment over weeks. A carrier whose trucks routinely get caught with them at roadside is a carrier where nobody is looking between breakdowns, and the roadside inspector has become the maintenance department.

A systematic operation runs preventive maintenance on intervals, and the intervals are enforced by something other than good intentions. Units get flagged by mileage or engine hours, a work order opens, the work gets done, and somebody verifies closure before the unit dispatches. The three questions that expose whether the system is real: what’s the PM interval, what happens when a unit blows through it, and who has the authority to hold a truck out of service against a load that’s already booked. That third one is the governance question wearing coveralls, and the answer predicts the out-of-service rate better than the out-of-service rate predicts itself.

Driver vehicle inspection reports close the loop, or they don’t. The regulation requires drivers to report defects and carriers to certify repair before the next dispatch when the defect affects safety. In a functioning program, DVIRs with defects trigger work orders, and you can trace a reported defect to a completed repair with dates that make sense. In a broken one, every DVIR in the building says no defects found, which isn’t evidence of good equipment. It’s evidence that drivers learned reporting defects gets them nothing but delay, and the carrier has built a system for not knowing about its own trucks.

For the broker or shipper who’ll never see any of this, the public proxies are the vehicle out-of-service rate against the national baseline, the violation mix skewing toward brakes and tires, and the trend over the trailing two years rather than the lifetime aggregate. A carrier trending down after a bad stretch is a different risk than one trending up from a clean baseline, and the trend is the part a snapshot check throws away.

12. Hours, ELDs, and the ghost co-driver

The electronic logging device mandate took effect in December 2017, and the industry story since has been that falsification ended when paper logs died. Falsification didn’t end. It professionalized.

The old fraud was a second paper logbook. The new frauds are built on how the ELD actually works, and you can’t spot them without knowing the mechanics. An ELD records driving time automatically off the engine, which is the part everyone understands. What it can’t do is verify who’s driving, and that gap produced the ghost co-driver: a second driver login, sometimes a real person who isn’t in the truck, sometimes an account invented from a compliant-looking CDL, that absorbs driving time so the actual driver’s record stays legal. One person drives fourteen hours, the log shows a team operation splitting it, and the paper is perfect. The tells are in the data if anybody looks: co-driver logins with no corresponding payroll, team operations where the second driver never appears on a roadside inspection, duty status changes that happen at highway speed.

The cruder versions are still everywhere. Unassigned drive time is the ELD’s honesty mechanism: the miles the device recorded with nobody logged in, and it’s supposed to be annotated and assigned. A carrier with chronic unassigned miles has drivers logging out to move the truck. Personal conveyance, the off-duty driving status meant for getting to a motel, has become the elastic category, with trucks moving hundreds of loaded miles in personal conveyance because the status doesn’t burn clock. Yard moves, malfunctions claimed at convenient moments, edits requested and approved by the same office login at midnight: every one of these lives in the ELD back office data, and none of it appears in any public record until it becomes a roadside violation.

What a carrier with real hours discipline looks like from outside: hours-of-service violations rare and clerical rather than substantive, no pattern of driving-after-hours or false-log findings, and, when you can see it, an operation whose transit commitments are physically achievable. That last one is a check brokers skip and shouldn’t. A carrier that accepts a 1,100-mile run for tomorrow morning with a solo driver has told you, at tender, that somebody is going to break the rule, and you accepted the answer. Dispatch math is the hours audit you can run from your desk, and after a crash, the transit time you demanded is in the plaintiff’s exhibit list either way.

13. Governance

Every program in this part reduces to one question: can the safety function tell the revenue function no, and survive it? Everything else is architecture around that answer.

You can find the answer in an org chart before you find it anywhere else. Where safety reports tells you what it is. A safety director reporting to operations is a compliance clerk, because the person who owns the freight owns them. A safety function reporting to ownership or a president, with documented authority to ground a truck, pull a driver, or refuse a load, is a control. Ask for the last time each of those authorities was actually used and what it cost. A carrier that can name the load it turned down last month has a safety program. A carrier that can’t has a safety manual.

Pay structure is governance written in numbers. A safety department bonused on revenue or on loads covered has been converted into a sales function with a different title. Drivers paid pure mileage with no detention or breakdown pay have been given a financial instruction to drive tired and skip the pre-trip, and no amount of posters in the break room outbids the pay plan. When I look at a compensation structure, I’m reading it as an instruction set, because that’s what it is. The company tells you what it wants by what it pays for, and the drivers heard it long before you asked.

Crash and claims discipline is the part an underwriter weighs heaviest, and the public record shows least. A carrier that does a real root cause review after an incident, preventable or not, and can show you what changed afterward, is running a learning loop. A carrier whose entire post-crash process is an insurance claim is running a payment loop. Ask what changed after the last serious incident. The good ones answer with a specific: a policy, a route, a piece of equipment, a termination. The rest answer with adjectives.

Size doesn’t decide this, and neither does sophistication. Some of the best-governed operations I’ve assessed were fifteen trucks with an owner who knew every driver’s family, held the line on maintenance because his name was on the door, and turned down freight that didn’t fit. Some of the worst had compliance departments, dashboards, and a safety culture deck, and a pay plan that unwound all of it every Friday. The org chart, the pay plan, and the last no. Those three tell you who’s actually driving the company, and the company that can’t answer them is being driven by whoever booked the next load.

Part IV. What the Losers Look Like

14. The chameleon

Same trucks, same people, new paper. That’s authority reincarnation, and the operator running the play is what the industry calls a chameleon carrier. The mechanics haven’t changed in twenty years because the incentive hasn’t: federal authority is cheap, history is expensive, and the system indexes history to the authority instead of to the people and the iron.

The play runs like this. A carrier accumulates violations, crashes, an intervention, a conditional rating, or an insurance market that won’t touch it at a survivable price. The principal forms a new entity, usually with a relative or an employee as the listed officer, obtains a new USDOT and MC number, moves the trucks and drivers over, and starts clean. The old authority dies by revocation or just goes dormant. The new one has no inspections, no crashes, no rating, and no history, and every compliance-only vetting tool in the market reads that absence as acceptable. The new entrant’s clean record is the product being manufactured.

FMCSA has known about this for a long time. A 2012 GAO report examined the problem and the agency’s ARCHI work, the algorithmic matching of new applicants against prior carriers, grew out of it, but enforcement capacity has never matched registration volume, and registration volume exploded after 2020. Screening for reincarnation is therefore a private diligence function, whether it should be or not, and it’s a solvable one, because the play leaves fingerprints at every layer of the anatomy from Part II.

The corporate layer: an entity formed weeks or months ago, holding fresh authority, at an address that housed another authority that died recently. Corporation younger than it has any reason to be, formation date against authority date against the death date of the neighbor. The people layer: officer names, and more usefully phone numbers and email addresses, shared with revoked or out-of-service carriers, because operators change company names far more often than they change cell phones. The iron layer: VINs inspected under the dead authority last year appearing under the new one this year, the physical fleet migrating across the paperwork. The service layer: BOC-3 process agent changes and clusters that track the same networks. No single tell convicts. Two or three together, on a new authority, aggressively soliciting freight, is a pattern, and the pattern is checkable before the first load in a way it never used to be.

The reason this section sits first in the losers part is that the chameleon defeats every other check you run. Its insurance is real, its authority is active, its record is clean, and every one of those facts is true of an entity that was constructed three months ago specifically so those facts would be true. Identity is the foundation check because until you know the carrier is who it says it is, and is the same operator it was last year, nothing else you verified means anything.

15. Ghost capacity, double-brokering, and the load that hauls itself

A carrier’s declared footprint is the most abused data in the industry, and the abuse has a purpose. Capacity that doesn’t exist is being sold every day, and the gap between what a carrier claims and what it physically runs is a leading indicator of fraud that almost nobody measures.

Start with the arithmetic nobody runs. A truck is good for roughly 100,000 to 120,000 miles a year run hard, which is two to three loaded long-haul moves a week sustained. A one-truck authority accepting forty loads a week isn’t a carrier. It’s a dispatch desk reselling your freight, and the reselling is the exposure, because the truck that actually shows up belongs to somebody nobody vetted. Power units against drivers against inspection volume against tendered volume: when those four numbers can’t describe the same physical operation, the operation you vetted isn’t the one moving the freight.

Double-brokering is the name for the resale, and the post-2020 version is industrialized. The classic version was a struggling carrier quietly re-brokering a load it couldn’t cover. The current version is organized: entities that obtain or purchase authority specifically to book freight and resell it, identity theft rings that impersonate legitimate carriers on load boards using spoofed emails and cloned MC numbers, and payment schemes where the fraudster collects from the broker while the actual hauling carrier, who thought it booked a legitimate load, never gets paid and liens the freight. Industry estimates put direct fraud losses in the hundreds of millions annually, and the number understates it because the liability exposure isn’t included. When the unvetted truck that actually hauled your load crashes, you selected that risk. You just did it blind, through an intermediary you didn’t know existed.

The controls are unglamorous, and they work at the point of physical custody, because paper identity is cheap and physical identity isn’t. Verify the contact channel against the carrier’s FMCSA-registered channel, not against the email that answered the posting, since the spoof lives in the reply-to. Confirm the truck, the driver name, and the unit number at dispatch, and match them at pickup. Watch the payment entity: a factoring assignment to a company that doesn’t match the carrier, a last-minute change in remittance, a carrier that’s oddly flexible about rate and oddly rigid about payment terms. Dormancy patterns matter here too. An authority that sat quiet for eighteen months and reactivated under new contacts with sudden volume is a shell that changed hands, and the record that made it look established belongs to a company that no longer exists in any meaningful sense.

The tell that costs nothing: the carrier that’s too easy. Available instantly for the hard lane, agreeable on rate, no questions about the freight, paperwork back in four minutes. Real trucks are scarce, and real carriers negotiate. A counterparty with frictionless everything is frictionless because it isn’t planning to do the part of the job that involves a truck.

16. The insurance tells

The industry checks insurance the way it checks a box: certificate on file, limits at a million, done. An underwriter reads the same file and sees six separate signals, and after a loss, a plaintiff’s lawyer reads it the way the underwriter does. This section is the underwriter’s read.

The certificate itself is the first misunderstanding. A certificate of insurance is a courtesy document generated by an agent, and it proves a policy existed at the moment of issuance, nothing more. The filing of record with FMCSA is better, and live verification at tender is the actual standard, because the failure mode is staleness: coverage that lapsed after the certificate was issued, which is precisely the scenario in which authority gets revoked for failure to maintain financial responsibility. A carrier in the window between insurance cancellation and authority revocation looks active and is functionally uninsured, and that window is where a measurable share of catastrophic uncovered losses live.

Lapse and reinstatement history is the pattern read. A carrier whose coverage has lapsed and reinstated three times in eighteen months is a carrier its own insurer keeps trying to shed, or a carrier that pays its premium only when the revocation notice arrives. Either way the market has priced this risk already and is telling you the answer, and insurer churn, a new carrier of record every renewal, says the same thing over a longer window. Insurance markets have better information about a carrier than you do. Watch what they do, not what the certificate says.

Who the insurer is matters as much as whether one exists. A policy from a thinly capitalized or unrated insurer, or from certain risk retention groups, is a recovery risk sitting behind a compliant filing. RRGs are a legitimate structure with a specific weakness: they’re capitalized by their member insureds, regulated by a single domicile state, and not backed by state guaranty funds, so when one fails, and several serving trucking have, its insureds are instantly uninsured, and its open claims are instantly unfunded. A carrier at a minimum limit, placed with a weak market, is telling you it bought the cheapest paper that satisfies the filing requirement, which is a statement about how it buys everything else too.

Structure and adequacy are the reads nobody does. A scheduled-auto policy covers the specific trucks listed on it, and a scheduled-driver endorsement covers the listed drivers, which means the substitute truck or the new hire on your load may be outside the coverage entirely, and you find out at the claim. Limit adequacy is the blunter problem: the federal minimum for general freight is $750,000, and the market convention is $1,000,000, both numbers set decades ago, and a single fatality routinely exceeds them severalfold. A million-dollar policy behind a carrier running dense urban corridors or high-verdict venues isn’t coverage against the loss that’s actually possible. It’s a deductible the plaintiff burns through on the way to the balance sheets behind it, and after Montgomery, one of those balance sheets is the party that made the selection.

17. The corporate tells

Exposure doesn’t respect the corporate boundary. Vetting tools do, and the gap between those two sentences is where sophisticated operators live.

The single-entity check is the design flaw. Every mainstream vetting product evaluates the DOT number in front of it, and the operators who matter run portfolios: multiple authorities under common control, formed at different times, with different listed officers, sharing equipment, drivers, addresses, and money. The portfolio exists because it works. It spreads history across entities so no single one accumulates enough to flag; it provides a fresh face when one burns, and it makes the do-not-use list a game of whack-a-mole, since flagging the entity does nothing about the operator.

Reading the network takes three public record types the freight industry mostly ignores. State corporate registries give you formation dates, registered agents, and officer names across entities, and the pattern of a person or an address appearing across multiple carrier registrations is the map. UCC financing statements are better than corporate filings because a lender filed them with money at stake: they name debtors and co-debtors, and a financing statement that lists the carrier, its principal personally, and two other trucking entities as co-debtors has just drawn the common-control diagram for you, sworn and dated. Litigation records complete it, because plaintiffs’ lawyers pierce these structures for a living and the complaints they file name the related entities and the principals in the first ten pages.

What you’re looking for in the network is traveling adverse history. A principal whose prior carrier was revoked, whose sibling entity carries a fraud judgment, whose related company sits on your own do-not-use list under a different DOT number. The entity in front of you can be six months old and spotless while the operator behind it has fifteen years of exactly the history you screen for, and the clean entity is the costume. This is the same logic as the chameleon check in section 14 run in the other direction: there you start from a suspicious carrier and look for the network; here you start from a clean one and check whether a network is what’s hiding behind it.

The boundary cuts both ways, which is worth saying because it keeps the check honest. Common ownership of multiple authorities isn’t itself adverse. Plenty of legitimate operators run separate entities for separate divisions, and a family with three trucking companies is often just a family with three trucking companies. The signal is adverse history plus the network, not the network alone, and a screening process that flags every multi-entity operator will bury the real finding under noise while disqualifying good capacity.

18. The carrier that looks fine and isn’t

The most dangerous profile in the data isn’t the ugly one. Ugly profiles get declined by everybody, including the tools this guide spends its time criticizing. The dangerous profile is clean, and it’s clean for a reason that has nothing to do with safety.

Thin data is the biggest category and the least understood. A carrier with two inspections has no record, not a good record, and the distinction is everything. Roughly speaking, the majority of active authorities are small enough and new enough that the public record can’t support a statistical conclusion about them in either direction, which means the majority of the market sits in a zone where a compliance-only screen returns nothing adverse as a matter of arithmetic. Nothing adverse reads as approval to a dispatcher under load pressure, and the system has just converted ignorance into confidence. The correct treatment is the opposite: thin data is a risk condition, scored as elevated until the record fills in, with the burden on verification and controls rather than on the absent history.

The new authority is thin data with a schedule attached. Carriers in their first eighteen months crash at materially higher rates than established ones, which is why authority age is among the strongest single predictors available, and the first eighteen months is precisely the period when the record is empty. The new entrant is simultaneously the least-known and highest-risk cohort on the board, and the market prices it backward, because new authorities buy freight with rate and the screen shows nothing adverse. None of this means new carriers are unusable. Every carrier was new once. It means the new one gets controls, defined freight, verification at pickup, and a file that documents you treated the unknown as unknown.

The stale rating is the clean look with a federal imprimatur. A satisfactory rating from a compliance review conducted a decade ago describes a company that may share nothing with the one in front of you but the DOT number: different owner, different fleet, different drivers, different everything. Ratings don’t expire, and most carriers are never reviewed twice, so the rating field on a profile is often the oldest fact on it, wearing the most authority. Read the date before you read the word.

The silent carrier is the subtle one. A fleet with the size and the tendered volume to generate steady inspections, showing almost none, isn’t lucky. Inspection exposure scales with miles, and a carrier moving real freight through weigh station states accumulates roadside contact as a matter of physics. Sustained silence from an active fleet means the exposure isn’t what’s claimed, the operation isn’t where it’s claimed, or the equipment isn’t crossing scales for a reason. A long quiet stretch from a carrier that’s supposedly running hard is a question, and the profile can’t answer it, which is the point. The clean profile ends the inquiry for a compliance check. For a risk check, it’s where the inquiry starts.

Parts V through VII follow: the vetting protocol step by step, scaling scrutiny to the load, defensibility, the deposition, and where the market goes from here.