The Carriers You Keep Loaded Are the Ones You Can Defend

I had a flatbed carrier I trusted more than most. MC-1174528 / DOT-3892104. Eleven months of active authority, zero crashes, vehicle OOS rate at 3.1% against a national average north of 20%, full cargo coverage current with a rated insurer, BASIC scores clean across the board. The kind of file you point to when you're explaining what a good carrier approval looks like.

He told me he was thinking about going elsewhere. Not because of rates. Because I kept sending him empty out of Laredo.

That conversation changed how I think about backhauls.

The Vetting Problem Nobody Names

The empty miles problem is usually framed as an economics problem — carrier pay, fuel burn, deadhead miles that cut into margin. It is that. But it's also a vetting problem, and most brokers don't see the connection until they're rebuilding a carrier pool from scratch.

When a carrier I've screened, monitored, and run consistently for six months leaves my rotation because I can't keep them loaded both ways, I have to replace them. And replacing them means starting over: new screen, new carrier file, new monitoring clock, a selection record that's one screen old instead of thirty loads deep. The documentation that took six months to build doesn't transfer to the next carrier.

Post-Montgomery v. Caribe Transport II, LLC — the Supreme Court's unanimous May 2026 decision that strips FAAAA preemption from state-law negligent-selection claims — this is a real dollar problem, not just an administrative one. A plaintiff's attorney isn't just looking at whether you screened the carrier before the first load. They're looking at the full record: How many times did you run them? What did monitoring show between loads? Did you have an ongoing documented basis for continued selection, or did you screen them once in October and hope nothing changed by March?

The depth of your carrier file grows with loads. A carrier you've run thirty-five times with a continuous monitoring log is defensible. A carrier you screened twice and used four times before a bad crash, with nothing in between, is not — even if the two screens were clean.

What 49 CFR § 371.3 Requires vs. What Actually Protects You

Brokers who know the reg at all know § 371.3. It requires a record for each individual transaction: the name of the carrier, their DOT number, the date of the transaction, the name of the consignor, the origin and destination of the shipment, and the amount of compensation received. That's the floor. The reg doesn't tell you to document why you selected this carrier over another, what you saw in monitoring between the first load and the twentieth, or that you re-ran their insurance at load tender every time.

The floor keeps you compliant. It doesn't keep you protected.

The documentation that actually matters in discovery is the selection rationale — not just "carrier has authority and insurance" but "carrier has authority and insurance, BASIC scores are within policy, no new crashes or violations since last screen, monitoring clear as of [date], prior twelve loads without incident." That record takes months to build and is impossible to retroactively create. The timestamp on every monitoring event, every screen, every carrier selection record is what either supports or undermines your testimony about what you knew and when you knew it.

None of that documentation exists on load one.

The File That Holds Up

When something goes wrong on load thirty-two with MC-1174528, my file looks like this:

• Initial screen, November of last year, full SAFER output, approval documented
• Thirty-one prior carrier selection records, each with authority status, insurance current, monitoring clear at time of tender
• Sixteen continuous monitoring events over eleven months, each with a note on what triggered the check and my review
• One monitoring flag in February — a Level 3 inspection in Odessa, tire pressure violation — reviewed and documented, no OOS, decision to continue use noted with rationale
• Two carrier agreement re-certifications, both timestamped, both before the loads that followed
• The Verifi truck/trailer confirmation on each load that included a pickup photo, trailer confirmation, and location stamp

That's a thick file. It shows a broker who selected a carrier thoughtfully, watched them consistently, and documented the basis for each selection decision — not once, but over eleven months of repeated use.

Now imagine a carrier I pulled off a load board last Tuesday because MC-1174528 couldn't cover the lane. I screened them that morning. They were clean. I ran them.

If something goes wrong on that load, my file is one screen from Tuesday morning. That's it.

Both carriers passed a screen. The exposure is completely different.

The Backhaul Math

Here's what keeping MC-1174528 loaded actually means in practice.

He runs Chicago to Laredo on a 48-foot flatbed. The lane is about 1,350 miles. If I'm paying $3.10 per loaded mile and he deadheads the return — or worse, takes a cheap backhaul from someone else at $1.60 per mile — he's losing money on the round trip at my volume. He starts shopping his availability elsewhere. I run him less. The gaps in my monitoring log start to look like neglect even though they're not. My documentation on him thins.

The difference between running a trusted carrier eighteen times a month and eight times a month is the difference between a rich, continuous file and a spotty one. The file quality doesn't track the screen quality — it tracks the load frequency. And load frequency, for a carrier who needs to run both ways to make the lane work, tracks backhaul availability.

If I can put a consistent backhaul in front of him on the return lane, he stays in my rotation, I run him more, and my documentation gets better with every load. The backhaul isn't just freight. It's ongoing file construction.

Finding the Freight Where the Truck Empties

I've been doing this manually for years. I'd call shippers I knew near the delivery point, check board postings for freight moving the right direction, sometimes work a trade — introduce a carrier to a shipper who could use them, hope the relationship stuck. It worked sometimes. A lot of times MC-1174528 went home empty anyway.

DOTScreener's Shipper Intelligence now has a backhaul finder that does this from the data instead of from memory. You put in where the truck empties and where it needs to go, and it shows shippers whose freight has actually been documented moving in that direction — filtered by equipment type. Flatbed, reefer, dry van, RGN. The data comes from FMCSA inspection records, so what you're looking at is observed freight patterns from real inspection activity, not load board posts that expired yesterday.

For a flatbed out of Laredo headed northeast, that's steel mills, ag equipment distributors, and construction suppliers near Laredo whose freight FMCSA inspectors have actually seen moving toward the Chicago market. If the direct lane is thin, it shows relay suggestions with documented freight on both legs. You can also pull up carriers who match the equipment if you're looking at it from the other direction.

I don't use it to generate a prospect list and blast sixty shippers. I use it to identify two or three shippers with consistent activity in the lane, call them directly, and build the kind of relationship where I can give a vetted carrier a reliable reload. The goal isn't volume. It's consistency for a carrier I've already decided is worth keeping.

How I Document This

When a carrier moves into my backhaul rotation, I add a dated note to their carrier file: the lane, the equipment, the shipper relationships that source the freight, and the date the arrangement started. That note is timestamped and it establishes continuity — it's the reason the thirty-fifth load happened, documented before the thirty-fifth load ran.

Every load in the rotation gets a carrier selection record: authority status, insurance current, monitoring clear, file note. Outbound and return, both. If I document only the outbound legs and something happens on a return run, I've created a gap on every backhaul load — and gaps get noticed.

If monitoring flags something between loads — a BASIC score crosses an alert threshold, an insurance filing lapses, a new crash entry appears — I review it before the next load and note the decision in the file. Not after. Before. The timestamp is what makes it evidence.

The file for MC-1174528 didn't happen because I was being unusually careful. It happened because I was running him consistently enough that the documentation accumulated naturally. The backhaul is why the load count was high enough for that to happen.

That's the thing most brokers miss. Good documentation isn't a separate process. It's what falls out of running fewer carriers more often, with a reason to keep them running both ways.

— Mason Lavallet

Founder, DOTScreener.com