AI Dispatch for Dry Van Carriers

Dry van is where the volume lives and where the margins die. A van carrier sees more available loads than any other segment, on more lanes, posted and re-posted all day. The problem was never finding a load. The problem is that there are too many to read, the rates are thin enough that a hundred deadhead miles or a two-hour detention turns a good load into a bad one, and the loads worth taking are gone in minutes. Dispatch for a van carrier is a speed-and-volume problem, and that is exactly the shape of problem software is good at.

Why dry van economics punish slow dispatch

Every carrier owns dry vans. The trailer is a commodity, so the freight prices like one. Brokers know any of a thousand carriers can cover a standard 53-foot van load, and that pricing power shows up in the rate you get offered. There were about 787,000 active carriers as of FMCSA data in December 2023, and per the ATA in 2025 roughly 91.5% of them run ten trucks or fewer. Most of those small fleets run vans, because vans are the cheapest equipment to buy and the easiest freight to find. That is the competition you are bidding against on every post.

The margin math is unforgiving in a way it is not for reefer or flatbed. ATRI's 2025 report put the marginal cost of operating a truck at about $2.26 per mile using 2024 data. A van carrier is often booking loads not far above that number, so the loaded rate alone does not tell you whether the load is any good. Deadhead does. Industry deadhead runs somewhere in the 15-30% range, and on a thin van rate, the empty miles to the pickup come straight out of the spread. A load that quotes well at the broker's number can be underwater once you add the repositioning to get to it and the likely empty miles after the drop.

Detention does the same kind of damage. The standard is two hours free before detention pay starts, and the industry loses an estimated $1.1-1.3 billion a year to it. On a high-rate specialized load you can absorb a few dead hours at a dock. On a $2.40-a-mile van load, an afternoon stuck at a shipper erodes the day's earnings and pushes the truck past the good loads on the next lane. Speed and positioning are not nice-to-haves in van freight. They are the whole margin.

The real bottleneck is reading the flood, not finding loads

A van dispatcher does not stare at an empty board. They stare at a full one. Pull up a busy outbound lane and there are dozens, sometimes hundreds, of postings, many of them the same load re-listed by different brokers at different numbers. A human reads them top to bottom, eyeballs the rate, guesses at the deadhead, and starts dialing. By the time they have worked through a screen of postings, the board has refreshed and the best one from five minutes ago is already covered.

This is where the volume turns against you. The dispatcher who can evaluate ten loads well will lose to the one who can evaluate two hundred quickly, because in van freight the winner is usually whoever calls first on a correctly-priced load. Manual triage caps how many loads a person can actually judge, and that cap is the ceiling on the fleet's revenue. The loads you never got to were not bad loads. You just ran out of attention before you reached them.

AI changes the order of operations. Instead of reading loads in the order a board happens to show them, the system scores the whole set at once on the number that actually matters: the all-in rate. It takes the posted line-haul, subtracts the deadhead to the pickup, factors fuel and the rough cost of the miles, and weighs where the load leaves the truck for the next move. Then it puts the loads that clear your threshold at the top. The dispatcher stops reading a firehose and starts working a short, ranked list of loads worth a phone call. That is the difference between reactive dispatch and dispatch that actually picks.

Ranking by all-in rate, not the number on the board

The posted rate is the most misleading number in van freight, and it is the one most dispatchers anchor to. Two loads can post at the same dollar-per-mile and have completely different economics once you account for what it costs to get to them and where they strand the truck afterward. Ranking by all-in rate is the single highest-leverage thing AI does for a van carrier, precisely because the segment's margins are too thin to survive booking on the sticker price.

Here is the kind of comparison the dispatcher should be making on every load, and rarely has time to:

Load B posts higher and runs less deadhead to the pickup, so reading the board top to bottom you take it. But the advantage can flip once you weigh the next move. Load A drops the truck where the next paying load is easy to find; Load B strands it somewhere you will burn empty miles to leave. Against the roughly $2.26-per-mile marginal cost in ATRI's 2025 report on 2024 data, both ends of that comparison are thin, and the destination is often what decides which one actually nets more over the next two loads. The all-in view catches that. The eyeball-the-sticker view does not. Across a month of loads, getting this right a few percent more often is the gap between a profitable van operation and one that runs hard for nothing.

None of this requires the carrier to chase a higher headline rate, which in a commoditized segment usually is not there to chase. The gain comes from stripping the loads that look fine and bleed money, and from moving first on the ones that genuinely clear. The broker still margins the load; DAT's 2023 data put gross broker margin around 13.5%, and that is not going away. What changes is that you stop being the carrier who takes the first number because the clock ran out.

Move first, then defend the rate

Ranking only helps if you act on it before the load is gone. In van freight that window is short, so the second thing AI does is compress the time between a good load appearing and a carrier making contact. The system can watch the boards continuously and surface or flag the loads that clear your threshold the moment they post, so the dispatcher is calling on the right load while it is still available instead of finding it after it is covered. That is a structural edge in a segment where being three minutes early is the whole game.

Negotiation is where the human stays firmly in control, and should. AI gives the dispatcher the floor to argue from: what this lane has been paying, what the all-in math needs to clear, where the posted number sits against recent rates. Walking into a broker call already knowing your number is worth more than any script. But the back-and-forth, the relationship with a broker who throws you repeat freight, the judgment call on whether to take a thinner load to keep a truck moving, those stay with the dispatcher. The tool does not own the phone. It hands the dispatcher a sharper starting position and gets out of the way.

A note on the same speed that helps you: it cuts both ways. Double-brokering and fraud are rising, with CargoNet reporting around $725 million in cargo theft in 2025. Moving fast on a load you have not vetted is its own risk. The point of ranking and surfacing is to spend your saved time on the calls that matter, including a quick check on whether the broker is real, not to fire off blind bookings faster. Speed is the edge; it is not a substitute for judgment, and the dispatcher is the one applying it.

Where the dispatcher stays in the loop

It is worth being plain about what AI does not do here. It does not replace the van dispatcher. The role is changing shape, not disappearing. Dispatching pay reflects real expertise; BLS put the 2023 median around $46,860, and the value of that person is going up, not down, because their time moves to the parts of the job that actually need a human. The software handles the reading and ranking that a person cannot do at the volume van freight demands. The person handles the brokers, the exceptions, the relationships, and the decisions that do not reduce to a number.

That division of labor is the whole pitch for van carriers specifically. In a segment defined by too many loads, too little margin, and too little time, the bottleneck is human attention, and that is the one thing you can hand off without losing control. The dispatcher still decides which load to book and at what rate. They just decide it from a short list of loads that already clear the math, instead of from whatever the board happened to show them before the clock ran out.

If you run vans and want to see this on your own lanes, the load-ranking-and-negotiation layer lives in AI Hub — it ranks the flood by all-in rate and keeps the dispatcher on the phone, in control of every booking.