Carrier Vetting After Montgomery - The Last Day

If you missed the first two installments they ran yesterday and the day before. Day one has a link to a comprehsive document thats downloadable.

So before we kick off the final installment of this series, I wanted to also give a shoutout to Attorney Brandon Wiseman for building a great carrier vetting program and training, which can be found here. While it is technically a Trucksafe product, I have nothing to do with it, and it was all Brandon. Check out his program at that link; it's interactive and at your own pace.

Carrier Vetting After Montgomery

Part V. How to Vet

19. The protocol

Everything in Parts I through IV converges on a sequence. The sequence is the same whether the seat belongs to a broker, a shipper, or an underwriter, because the carrier and the risk are the same. Only the depth changes with the seat and the freight. What follows is the order, why the order is the order, and how each step fails when it’s skipped or run out of sequence.

Identity comes first, before anything else is trusted. Every downstream check verifies facts about an entity, and if the entity in front of you isn’t the entity those facts describe, you’ve verified somebody else. Confirm the legal name, DOT and MC together, the FMCSA-registered contact channel rather than the channel that answered you, and continuity: same operator this year as last year, same iron, no migration pattern. A fluid identity doesn’t get scored. It stops the process, because you can’t underwrite what you can’t identify, and every chameleon and double-brokering loss in the record started with an identity somebody took on faith.

Then the eligibility floor, live, not cached. Active authority for the transportation being arranged, no out-of-service order, required financial responsibility in force and applicable. These are pulled at the time of the decision from the live federal record, because authority status and OOS orders change daily and a cached snapshot fails in both directions: it rejects the carrier whose order was lifted Tuesday and passes the one revoked Thursday. The floor items aren’t judgment calls and nobody in the revenue chain gets to waive one because the load is hot. A salesperson can’t clear a hard stop. That sentence belongs in your written policy in roughly those words.

Then measure the exposure, which is where the compliance check ends and the risk check starts. Read the safety record as outcomes: crash rate against exposure rather than raw count, out-of-service results against the national baseline, severity over paperwork, trend over snapshot, recency weighted over ancient history. Then read the signals that live outside the FMCSA record entirely, because they’re what a plaintiff pulls and a compliance tool doesn’t: federal litigation history and how recent and how severe, corporate structure and the network behind the authority, insurer identity and policy structure rather than filing presence, and the coherence of the declared operation, trucks against drivers against inspections against the volume the carrier is accepting. A carrier that fails coherence isn’t a scoring question. It’s a section 15 question.

Then scale it to the load, which section 20 covers, because the same carrier is a different total risk under different freight.

Then read confidence and trajectory before you read the score. How much record actually stands behind the number, and which direction the carrier is moving. A thin record is elevated risk, not clean risk. A carrier trending down out of a bad stretch is a different decision than the same score trending up, and the trend is the fact a snapshot throws away.

Then document the decision, at the time of the decision. Section 22 is that step in full. The short version: the file gets written when you say yes, in risk terms, with the data as it stood that day, or the file effectively doesn’t exist.

Then monitor and re-vet on triggers, because selection decays. Authority revocations, insurance cancellations, out-of-service orders, crashes, and score deterioration all arrive after onboarding, and a vetting program that runs once has a shelf life measured in weeks. The triggers that matter are the fast ones: revocation, lapse, OOS order, reportable crash. The cadence items are the slow ones: rating changes, trend deterioration, network changes behind the entity. A carrier unused for ninety days gets re-verified at tender as if it were new, because for your purposes it is.

The protocol reads long and runs short. Steps one and two are seconds with the right tooling. Step three is the depth dial, thirty seconds for routine freight and thirty minutes for the freight that deserves it. The sequence is the part that can’t flex, because each step assumes the ones before it, and the industry’s standard failure is running step three carefully on an identity nobody confirmed in step one.

20. Scaling scrutiny to the load

A vetting standard that treats every load the same is wrong twice. It over-inspects the routine freight, which trains your people to route around the process, and it under-inspects the freight that can end the company, which is the failure you actually can’t afford. Risk is a property of the carrier and the load together, and the standard has to price both.

Severity scales with the freight in ways that are knowable at tender. Hazmat carries release exposure and a federal incident history you can check, and it converts a crash into an evacuation. High-value and theft-targeted commodities, electronics, pharmaceuticals, alcohol, copper, pull organized theft the way a porch light pulls moths, and the theft rings work the same load boards you do. Dense urban lanes put the truck next to more people per mile. Venue matters and pretending it doesn’t is malpractice with extra steps: the same crash produces materially different verdicts in different jurisdictions, the plaintiff bar publishes its own venue rankings, and a lane running through a high-verdict county is carrying that fact whether your process prices it or not. Team and expedited service adds fatigue exposure by design, because the service commitment is the hours pressure. Each of those facts is visible before the freight moves, and each one should tighten the screen: higher score floor, senior approval, verified rather than filed insurance at limits matched to the exposure instead of the 1980s minimum, tracking, pickup verification, and a documented file.

The insurance piece deserves its own sentence because it’s where load scaling gets concrete. A million-dollar limit behind a dry van of paper towels on a rural lane and the same limit behind a tanker through a metro corridor aren’t the same coverage. The required limit for the load is a function of what the load can do, and a carrier whose coverage doesn’t answer the freight’s real exposure is effectively uninsured for your purposes even though its filing is perfectly compliant. That’s an eligibility question, not a scoring question, once the gap gets big enough.

The other half of scaling is the improvement direction, and it’s the half that keeps the standard from strangling your capacity. A standard that only declines produces a carrier pool of incumbents and starves everything else, which is commercially impossible and also bad risk policy, since the marginal carrier that’s genuinely improving is tomorrow’s core carrier. The answer is graduated use: the not-yet-proven carrier gets defined freight, defined lanes, trial volume, controls at pickup, and monitoring, and earns its way up as the record fills in. That graduated file, this carrier was new, we treated it as new, here were the controls, is itself a defensible document. Blanket exclusion isn’t the defensible position. Proportionate response is.

21. Reading the data the way a courtroom reads it

The same public record gets read three ways: by a dispatcher deciding in ninety seconds, by an underwriter pricing a book, and by a plaintiff’s expert building a case. The third read is the one your file eventually gets, so you should know what it looks like, and the honest way to know is to describe the workup, because it’s the same workup I’m retained to perform.

The expert starts with the crash and works backward to the selection. The carrier’s profile gets pulled as of the tender date, not today’s date, because the question is what you could have known when you said yes. The Freedom of Information Act reaches the historical MCMIS snapshots, so the profile as it stood that Tuesday is recoverable whether you saved it or not. Then the record gets read for everything Parts II and IV cover: the rating and its date, the OOS rates against baseline, the violation mix and what it says about maintenance and hours culture, the crash history with severity, authority age, the corporate formation date against the authority date, officer and address overlaps, the insurance history with its lapses, and the litigation record. Every red flag that existed on tender day goes in a column, and the column becomes a demonstrative titled something like what the defendant could have seen for free.

Then your process gets read against the column. The requests for production ask for your carrier selection policy, your vetting file for this carrier, your monitoring alerts, your exception approvals, and your communications about this load. The depositions ask who checked what, when, and what they did with it. The two failure modes are mirror images. If you had no process, the argument is that a reasonable company looks and you didn’t. If you had a process and didn’t follow it, the argument is worse, because your own policy becomes the standard of care and your own document proves you breached it. A written standard you honor is a shield. A written standard you ignore is the plaintiff’s best exhibit, and they’d rather have it than nothing.

The defense read is the same data run the other direction, and it’s worth understanding because it’s what a good file makes possible. The defense expert shows what the profile looked like on tender day: the rating that was satisfactory or unrated, the OOS rates that sat at or under baseline, the crash history that was unremarkable at the carrier’s size, the flags that existed and the controls that answered them, the documented reasoning. Hindsight is the plaintiff’s whole advantage, and a contemporaneous file is the only thing that takes it away, because it moves the argument from what the data shows now to what a reasonable company did then. That’s the entire game, and it’s decided before the crash by whether the file exists.

Part VI. Defensibility

22. The file is the case

After a catastrophe, the selection file is the case. Not a piece of it. The case. Liability on the carrier is usually clear, and damages are usually enormous, so the contested question, the one the years of litigation are actually about, is whether the party that chose the carrier was reasonable to choose it. That question gets answered by a file, and the file either exists from the day of the decision or it gets reconstructed two years later by lawyers, from fragments, under oath, badly.

Contemporaneous is the whole word. The federal data changes continuously, so the carrier’s profile at deposition time can look nothing like it did at tender, and without a dated record the jury sees today’s ugly profile with the plaintiff’s implication that you saw it too. A dated snapshot of what the record actually showed on the day you said yes is the single most valuable document in the defense, and it’s also the cheapest, because capturing it costs a timestamp and storage. The gap between those two facts, decisive value and trivial cost, is the strongest argument in this entire guide, and it’s the one I’d put on a poster in every brokerage in the country.

What the file contains, for any carrier that isn’t routine, and in risk terms throughout. The identity as verified: legal name, DOT and MC, the contact channel used and how it was confirmed. The eligibility items as pulled live, with the timestamp. The data as it stood: the rating and its date, the safety figures, the insurance verification with insurer and limits and structure, and any flag that existed. The load and its severity profile, because the reasonableness of the selection is judged against the freight. The controls applied: tracking, pickup verification, direct insurance confirmation, trial limits, whatever answered the flags. The reason for proceeding, stated as risk reasoning. The approver, their role, and the scope of the approval. If the carrier was on a graduated or improvement track, include the plan and the checkpoint status, because a carrier under documented monitoring is materially easier to defend than the same carrier chosen cold.

Build it into the workflow, or it won’t happen. A documentation step that requires a dispatcher to assemble screenshots after the fact will be skipped exactly as often as the dispatcher is busy, which is always, and inconsistent documentation is its own exhibit, since the loads with files make the loads without files look like decisions to hide something. The capture has to be automatic, attached to the tender event, and exportable on demand. The test of the system is blunt: pull any load at random from two years ago and produce, in minutes, a dated file showing what was checked, what it showed, and why the answer was yes. If your operation can’t pass that test today, that’s the gap to close before any other spending in this guide.

23. What the record says, and what it never says

Discovery produces two kinds of documents: the file you built on purpose and the communications you didn’t think of as documents. Emails, texts, TMS notes, and chat messages all get produced, and the sentences your people write under load pressure are the sentences a jury hears read aloud. This section is about those sentences.

The ones that end cases are ordinary. Needed the truck. Customer was screaming. Rep’s used this guy forever, he’s fine. Just get it covered. They’ve got authority and insurance, send it. Every one of those is a normal operational sentence, typed a thousand times a day across the industry, and every one of them converts, in a courtroom, into an admission that capacity outranked safety at the moment of decision. The improper-hiring theory that drives verdict size runs entirely on sentences like these, because they let the plaintiff argue the choice was knowing. The crash was an accident. The email is a decision, and juries punish decisions.

The fix isn’t coaching people to write carefully, which fails, and it definitely isn’t telling people to stop writing things down, which is spoliation bait and worse than the disease. The fix is giving the pressure somewhere legitimate to go. When a dispatcher needs an exception, the process routes it to someone with risk authority and no commission on the load, and that person’s documented reasoning is the record: the flag, the mitigation, the scope, the approval. The exception happened either way. The difference is whether the record is a reasoned risk decision by an authorized person or a text that says, “Just get it covered.” Same load, same carrier, opposite exhibits.

Approval authority is the structural version of the same point. The person who clears a flagged carrier can’t be a person paid on the load, because the plaintiff’s cross writes itself: you earned commission on this shipment, and you also decided the safety concern didn’t matter. Risk approval sits with someone whose compensation doesn’t move with the answer, and that org chart fact, cheap to implement, testifies well, and its absence testifies terribly, is one of the highest-leverage changes in this whole part.

24. Contracts, indemnity, and the empty pocket

Selection diligence governs whether you chose well. Contracts govern who pays when something goes wrong anyway, and the standard broker-carrier agreement in circulation was drafted for the world where preemption did the heavy lifting. Post-Montgomery, the paper needs to actually work, and whether it works comes down to three structural points and one hard truth.

The insurance provisions are the load-bearing wall. The agreement specifies minimum coverage by type and limit, requires the carrier’s policy to be primary and non-contributory so the carrier’s insurer answers before yours, and names you as additional insured where the structure supports it. Each phrase does a job. Primary and non-contributory is the difference between the carrier’s insurer defending the claim and a coverage fight where your own carrier pays first and chases later. Additional insured status gets you a defense under their policy rather than just a reimbursement promise. None of it means anything if the verification in the protocol didn’t happen, because the contract can require a million any-auto and the carrier can be carrying scheduled coverage from a sinking RRG, and you find out which one at the claim.

The re-brokering prohibition is the contractual half of section 15. The agreement bars re-brokering, interlining, or substituting another carrier without written consent, and states that the carrier accepting the load is the carrier physically hauling it. The clause doesn’t stop a fraudster; paper never does. What it does is establish the breach cleanly, support the claim against the carrier that handed off your freight, and, just as importantly, document that your program prohibited the thing, which matters when the question is whether you were reasonable.

Indemnification runs in your favor and covers the carrier’s negligence, with the carrier obligated to defend as well as indemnify. Then the hard truth, and it’s the sentence this section exists for: an indemnity clause is only as good as the balance sheet behind it. A promise to make you whole from a three-truck carrier with a minimum-limit policy from a weak market is a promise from an empty pocket, and when the pocket is empty, the plaintiff doesn’t go home. The plaintiff comes to you, which is exactly the scenario Montgomery reopened. That’s why the contract section of this guide is shorter than the vetting sections. The paper allocates the loss. Only the selection prevents it, and only real, verified, adequate coverage makes the allocation collectible. Get the terms reviewed by transportation counsel against the post-Montgomery landscape, and then remember that the best contract in the industry has never stopped a truck.

25. The deposition

The deposition is where the program you actually ran gets separated from the program you describe, and it follows a structure you can prepare for years in advance, because the structure is the same in nearly every case. What follows is the sequence as it runs, and what each phase is actually testing.

It opens with the witness’s role and the company’s process in the abstract, and the questions sound easy because they’re supposed to. Walk me through how a carrier gets approved. Who’s responsible for vetting? What does your policy require? The phase feels like background, and it’s actually the trap being set, because everything the witness claims here becomes the standard the specific load gets measured against. A witness who describes an aspirational process, the one in the manual rather than the one on the floor, has just built the gap the rest of the deposition will live in.

Then the policy documents come out, and the questions get specific. Your policy requires X. Was X done on this load? Show me where. The exhibits are your own documents: the selection policy, the vetting file or its absence, the TMS records, the emails from section 23. Every gap between the described process and the documented reality gets walked through one item at a time, slowly, on the record. This phase is why section 22 says the file is the case. A witness holding a dated, complete file answers these questions in minutes and the phase collapses. A witness without one spends three hours explaining absences, and every I don’t recall is a brick in the closing argument.

Then hindsight arrives dressed as fairness. The carrier’s profile from tender day goes on the screen, or worse, today’s profile if you can’t establish the tender-day version. You could see the conditional rating, correct. You could see the out-of-service rate, correct. This information was free, correct. You looked at none of it, correct. Four yeses in a row, each one individually harmless and honest, and the sequence is the case. The only answer that survives that sequence was written two years earlier: yes, we saw it, here’s the file, here’s what we did about it, here’s the reasoning, here’s who approved it. Defensibility isn’t eloquence in the chair. It’s the document that makes the chair boring.

The witness who does well is almost never the one who talks well. It’s the one whose answers are short, true, and backed by paper, who says I don’t know when they don’t know instead of improvising, and who never argues, because arguing hands over sentences. Preparation for that witness starts at tender, not at the subpoena. Every practice in this guide, the live checks, the file, the exception routing, the approval authority, exists so that the person in the chair can answer the hardest question in the case: why was this carrier a reasonable choice, by sliding a document across the table. Companies that can do that settle well or win. Companies that can’t fund the verdicts everyone else reads about.

Part VII. Where This Goes

26. The market sort

Montgomery doesn’t regulate anybody. What it does is reprice information, and repriced information sorts markets. The sort is already visible, and it runs through three channels.

Insurance moves first because insurance always moves first. Contingent auto and broker E&O were priced against a world where preemption killed most negligent selection claims before discovery, and that assumption died on May 14. Underwriters reprice by asking questions, so broker renewals now come with questionnaires about carrier selection standards, documentation, monitoring, and exception governance, and the answers move premium. A brokerage with a real program becomes a better risk than its peers for the first time in a way that shows up in dollars, which is the mechanism by which everything in Parts V and VI stops being a cost center. The market is about to pay you for your file.

Freight follows insurance. Shippers who understand section 10’s second-order exposure start auditing their brokers’ selection processes, because we relied on our broker only works as a defense if the broker was worth relying on, and proving that requires evidence gathered before the crash. Broker RFPs grow a diligence section. Brokers, in turn, steer volume toward carriers who can prove how they operate, because a carrier that can hand over its own operational evidence, real hours discipline, real maintenance closure, real driver management, makes the broker’s file stronger. The carriers who can show their work get the freight, and the demand for provable operations flows down the chain to the telematics and compliance infrastructure that produces the proof.

The bottom of the market contracts, and honesty requires saying what that means. The carriers that lose in the sort aren’t only the frauds. They include the marginal-but-real operators whose records can’t survive scrutiny and whose freight access tightens as the screens tighten. Some of that is exactly the point: capacity that shouldn’t exist exiting the market. Some of it’s the improvement problem from section 20, real carriers that need a route up rather than a wall, which is why graduated use and documented improvement matter as market policy and not just as legal defense. A sort with no ladder just pushes the bottom of the market further into the shadows, and the shadows are where the chameleons already live.

27. Rulemaking and the federal picture

The private-market sort runs ahead of the government, which is normal, but the federal track matters because it will eventually define terms everyone else has to use. Three threads are worth watching, and one honest caveat applies to all of them: rulemaking timelines slip, administrations change priorities, and anything in this section describes direction, not schedule.

The safety fitness determination problem is the oldest thread. The current three-tier rating system, satisfactory, conditional, unsatisfactory, reaches only the small fraction of carriers that receive compliance reviews, which is why most of the industry is unrated and why the rated field on a profile is so often a decade old. FMCSA has worked toward a revised safety fitness methodology for years, including the abandoned 2016 proposal and the renewed effort announced in the 2020s, and the core question has never changed: whether fitness can be determined from data rather than from site visits the agency lacks the staff to perform. Any rule that emerges will effectively federalize a version of the outcome-based reading this guide describes, and the companies already reading carriers that way will find the transition trivial while the checkbox operators rebuild.

The new entrant thread is the one closest to the problems in Part IV. The current new entrant program grants authority first and audits inside eighteen months, which is precisely backward for the chameleon problem, since the fraud is committed at registration and the audit arrives after the freight has moved. The anticipated direction is more scrutiny at the front gate: identity verification, fitness screening before authority rather than after, and harder matching of applicants against prior operators. Whatever final form that takes, the diligence burden it implies is one the private market is already carrying, because nobody selecting carriers today can wait for a rulemaking to solve the new entrant cohort for them.

The credit-for-safety thread is the sleeper. The FAST Act directed FMCSA to establish a Beyond Compliance program recognizing carriers that invest in safety technology and practices above the regulatory minimum, and the concept has mostly sat dormant at the federal level while the market built it anyway: insurers pricing telematics, brokers preferring carriers with provable operations, the whole show-your-work economy of section 26. If the federal version ever matures, it converges with the market version, and the direction of both is the same sentence this guide keeps arriving at. The floor is compliance. The question is what a carrier can prove beyond it.

28. What I’d do Monday if I ran a brokerage

Everything above compresses into a ninety-day sequence, ordered by leverage, with the parts that cost attention separated from the parts that cost money. Most of the leverage is in the attention half, which is the fact that should change how you read the whole list.

The first two weeks cost attention. Write the carrier selection standard, or rewrite the one nobody follows, and make it the standard you’ll actually run: floor items, score thresholds, load scaling, exception routing, approval authority held by someone with no commission on the load. Move approval authority now, because it’s an org chart edit with the testimonial value of a capital project. Kill the cached status checks and make authority, OOS, and insurance verification live at tender. Send the two sentences that cost nothing and change behavior immediately: sales can’t clear a hard stop, and exceptions go to risk in writing.

The first month closes the documentation gap, because section 22’s random-load test is the whole program in one question. Whatever your systems are, the tender event has to capture a dated snapshot: what was checked, what it showed, who approved it, and why. If your TMS can’t, bolt it on, and if you have to choose between spending here and spending anywhere else in this list, spend here. The file is the case, and the file is built at tender or never.

The first quarter builds the depth. Stand up monitoring with real triggers, revocation, lapse, OOS, crash, and a documented response path, because an alert you didn’t act on is worse in discovery than no alert. Run your active carrier list through the Part IV screens: identity, network, coherence, insurance structure, and expect findings, because every book has them. Build the graduated track for new and marginal carriers so declines become pathways and your capacity survives your standards. Pull your broker-carrier agreement and put it in front of transportation counsel against the post-Montgomery landscape. Then audit yourself before anyone else does: pull twenty random loads, run the test, and fix what fails.

None of this is beyond a ten-person brokerage, and none of it’s optional for a thousand-person one. The economics changed on May 14 in one specific way: the work this guide describes used to be a cost you could skip because preemption made the skipping free, and now the skipping is the most expensive decision in the building. The data was always public. The carriers were always knowable. The only thing Montgomery changed is that somebody finally gets to ask, in front of a jury, whether you looked. Make the answer yes, and make it a document.