Your Carrier Policy Doesn't Protect You. Your Per-Load Record Does.

The deposition went sideways around 11:15 in the morning. Not because the broker had a bad carrier vetting policy — she had a good one, actually. Clear, specific, signed off by counsel. Checked OOS rates, authority age, insurance. The works.

The plaintiff's attorney put a piece of paper on the table. "This is the carrier packet for MC-1247893, approved by your company in February of 2024. I'm not asking about the policy. I'm asking about load number 4-7-8-2 on October 9th, 2024. Show me the record that proves you checked this carrier's safety status before you tendered that load."

Silence.

The broker checked carriers at onboarding. She had a clean file for that carrier from eight months earlier. She hadn't checked again before that specific load. Neither had anyone on her team. The carrier's Unsafe Driving BASIC had moved from the 40th to the 74th percentile in the window between approval and tender. Nobody caught it.

That's not a policy failure. That's a per-load failure. And after Montgomery v. Caribe Transport II, LLC made negligent selection claims actionable in every state court in the country, the per-load failure is the one that puts you on the wrong end of a verdict.

The policy is a floor, not a ceiling

Your written carrier selection policy tells a jury what your standards are. The per-load record tells them whether you actually met those standards on the day in question. A policy without a per-load record is a statement of intent. It documents what you were supposed to do, not what you did.

Think about how a negligent selection case actually works at trial. The plaintiff's expert gets up and walks the jury through your policy — "Step 3 says verify current OOS rate before tender" — and then shows them discovery where no such verification exists for the load in question. The policy just became the instrument of your own conviction. You published your standard and then proved you didn't meet it.

A good per-load record does the opposite. It creates a contemporaneous log — timestamped, specific, tied to that load number — showing what you checked, when, and what you found. It doesn't have to be elaborate. It has to exist.

What "per-load" actually means in practice

You don't re-onboard a carrier from scratch every time you use them. That's not what per-load means. What it means is that before you tender a load to a carrier, you take a fresh look at the safety data that can change between onboardings — and you log it.

The things that can change quickly: OOS rate (updates monthly in the Safety Measurement System), BASIC percentiles (update monthly), authority status (can be revoked at any time), insurance status (can lapse with 30-35 days notice under BMC-91 or no notice at all for some surplus-lines endorsements). The things that change slowly but are still worth a spot-check: safety rating, complaint history, crash count.

For most loads, a full re-screen takes three minutes. For loads above $100K value, high-value commodities, or anything that takes the carrier through a weigh station corridor with an already-elevated OOS rate, add two more minutes to look at the crash detail.

Here's what I actually run before confirming any load:

FMCSA SAFER snapshot check. Current authority status — I need "ACTIVE," not "OUT OF OPERATIVE." I note the date and time. If it says anything else, the load doesn't move until I understand why.

Insurance verification. Active BMC-91 or BMC-91X on file, effective date, expiration date if showing. Cargo policy: limit and policy number. A $750K BIPD minimum is what 49 CFR § 387.9 requires for general freight, but on a pharmaceutical or electronics lane I'm looking for $1M cargo minimum and I'll ask for the full ACORD 25 if the load value is above $200K. I note what I saw and when.

OOS rate quick-read. I'm not recalculating it — I'm pulling the current percentile from SMS and logging it. If it moved more than 15 points since the last time I used this carrier, that's a conversation before I click confirm.

BASIC flag scan. Takes 20 seconds. I'm looking for any BASIC that's crossed into alert (above the intervention threshold) since the last tender. If the Unsafe Driving BASIC is flagging — above 65th percentile for large carriers, above 75th for small — that's something I document a decision on. Not necessarily a no, but a documented choice with a reason.

Authority age confirmation. This matters most the first two or three times I use a carrier that's under 36 months old. After that it fades into background context.

That's the base run. Five minutes. The output is a short entry — three to five lines — in the load file: date/time, what I checked, what the numbers were, any flags and my decision on them.

The specific fields that show up in discovery

Plaintiff's attorneys in trucking cases aren't guessing what to subpoena. After a few years of post-Montgomery litigation, they have standardized discovery requests that ask for every broker-generated document associated with a specific load. What they're looking for, specifically, is the per-load check record — and when it doesn't exist, that absence is its own evidence.

Here's what I've seen come up repeatedly in the context of these cases:

The SMS pull date. If you can produce a screenshot or a log entry showing you checked the carrier's BASIC scores on the day of or day before tender, that's contemporaneous evidence. A note that says "checked carrier SAFER 10/9/24 at 9:47 AM — Unsafe Driving 41st percentile, no BASICs in alert" takes 30 seconds to create and takes a key argument off the table.

The insurance check timestamp. Under 49 CFR § 387.9, the carrier is required to have the minimum financial responsibility on file. You checking it doesn't fulfill their legal obligation — but it shows your diligence. A verified insurance check with the date becomes part of your defense.

The OOS rate at time of tender. One of the arguments plaintiff's attorneys make is that a simple check of public data would have revealed the carrier was unsafe. If your log shows you checked the OOS rate that day and it was 6.2% on 16 inspections — completely normal — that argument fails on the data.

The decision log when something was flagged. This one is counterintuitive. If you checked a carrier and found an elevated BASIC and tendered the load anyway, that can look bad — unless you logged your reasoning. A 68th-percentile Unsafe Driving BASIC on a 3-inspection dataset is very different from a 68th-percentile on a 30-inspection dataset. If you saw that, thought about it, decided the small inspection count made the percentile unreliable, and wrote that down — that's a documented exercise of professional judgment. It's defensible. The same action with no documentation is just "we ignored the flag."

When to run the deeper check

Most loads don't need more than the base five-minute run. But certain triggers should put you on a longer protocol:

A carrier you haven't used in more than 90 days. A lot can change in 90 days. Run it like a fresh approval.

A load over $200K in cargo value, or anything with PII, pharmaceuticals, electronics, or other high-theft commodities. The cargo value means the stakes are higher; the commodity type means cargo insurance exclusions matter and you should pull the full policy detail.

A carrier whose BASIC scores have moved since your last use. If the Unsafe Driving BASIC went from 38th to 71st percentile in the last 60 days, you need to understand why before you tender.

Any carrier with a new DOT crash record since your last use. Even if the crash is coded "non-preventable" — that designation doesn't appear in real time. The crash appears first; the preventability determination comes later, sometimes months later.

A carrier that just acquired another carrier's authority or changed ownership. This one trips a lot of people up. The SAFER snapshot will show a clean new-entrant record but old equipment; the underlying safety culture might be completely different from what their fresh start suggests. I treat ownership changes like a fresh approval with extra scrutiny.

How I document this

My per-load log is a running entry in my TMS notes field, supplemented by a screenshot-to-folder habit for anything that shows a flag or a decision point.

For a standard load, it looks like this: "Vetting check 10/9/24 9:47 AM CT: DOT-3567102 / MC-1247893 authority active, insurance current (BMC-91 on file, cargo $1M, exp. 3/2025), Unsafe Driving 41st pct, no BASICs in alert, OOS 6.2% on 16 inspections, authority 22 months. Proceed."

For a load where I found something and made a decision: "Vetting check 10/9/24 9:47 AM CT: DOT-3567102 authority active, insurance current. HOS BASIC 71st pct — 4 inspections, low data reliability, discounted per small-n guidance. Unsafe Driving 38th pct. OOS 8.1% on 22 inspections, slightly elevated but within acceptable range for lane. No crash additions since 8/12 last use. Proceed, flagged in file."

If I took a screenshot of the SAFER snapshot or the SMS data, it goes into a folder named by DOT number and date and gets attached to the load record. Thirty seconds of housekeeping per load.

Three years from now, if a plaintiff's attorney asks what I checked before I tendered a load to DOT-3567102, I can pull up that file and hand it to them. The check is documented, the decision is documented, and the carrier's actual status at time of tender is in the record. That's the game.

Why this matters more than it used to

Before Montgomery v. Caribe Transport II, LLC, the preemption argument was doing a lot of heavy lifting for brokers. The 7th and 11th Circuit precedent held that the FAAAA preempted state negligent selection claims — which meant that even if a broker's vetting was thin, there was a ceiling on state-court exposure. That ceiling is gone.

The unanimous Supreme Court opinion in May 2026 means every state court plaintiff's bar now has a clear path to sue brokers for negligent selection. The ones who win those cases will be the ones where the broker's file is thin, the per-load check is absent, and the carrier's safety signals were there for anyone paying attention. The ones who get dismissed or settle small will be the ones who can show a contemporaneous record of reasonable care.

The difference is usually a five-minute habit. Not a new system, not a new tool, not a new policy. A five-minute habit, logged somewhere permanent, with a timestamp.

How I document this

At load tender:

• TMS notes entry: carrier name, DOT/MC, date/time, authority status, insurance status with expiry, top-line BASIC percentiles, OOS rate with inspection count, authority age, and outcome (proceed / flag / decline)
• Screenshots of SMS data when any BASIC is above 55th percentile or OOS rate is above 8%
• Folder structure: one folder per DOT number, subfolders by year, screenshots named by load ID

For flagged loads:

• Reason for flag, reasoning for decision, who approved it if elevated

Retention: I keep everything for at least 5 years. The statute of limitations on personal injury claims in most states is 2-4 years, but discovery starts from the complaint date, not the incident date.

This isn't the expensive part of running a brokerage. It's the part most people skip because it feels like overhead until the day it isn't.

— Mason Lavallet

Founder, DOTScreener.com