Dispatch Software for 20-Truck Fleets: When Manual Breaks

At five trucks, dispatch lives in one person's head and a spreadsheet, and that is enough. At around 20, it stops being enough. The loads slip, the check calls get missed, the deadhead creeps up, and nobody can answer "where is everything right now" without three phone calls. The work did not get harder; it got wider than one person can hold. This post is about recognizing that breakpoint and deciding what to automate first when you reach it.

Why 20 is the number where it breaks

There is nothing magic about 20 specifically, but the math around it is consistent. A single dispatcher can carry a handful of trucks in working memory: which driver is where, who is empty Thursday, which broker still owes a rate con. The spreadsheet is a backup, not the system of record. It works because the whole operation fits inside one skull. Somewhere in the teens, that stops being true. The number of moving parts crosses the line where no one can hold all of them at once, and the spreadsheet that was a backup is suddenly the only thing standing between you and a missed pickup.

For context on where a 20-truck fleet sits: of roughly 787,000 motor carriers in the US (FMCSA, December 2023), about 91.5% run 10 trucks or fewer and 99.3% run fewer than 100 (ATA, 2025). A 20-truck fleet is already past most of the field. You are not a tiny operation anymore, but you are also nowhere near the size that justifies a full back-office team. That is exactly the awkward middle where the manual approach has run out of road and the obvious fix, hiring, looks expensive.

And hiring is expensive. The median dispatcher earns about $46,860 a year, roughly $22.53 an hour (BLS, 2023), before you count the ramp time, the desk, the software seat, and the months it takes a new hire to learn your lanes and your brokers. That is real money against margins that are already thin. So the honest question at 20 trucks is not "should I add a person." It is "what is actually breaking, and can I fix that part without adding a person yet."

What actually breaks at 20 trucks

The failures are specific, and they tend to arrive in roughly this order:

• Loads slip through the gaps. A good load comes across a board or a broker email while the dispatcher is on the phone about a different truck. By the time anyone looks, it is gone or the rate dropped. At five trucks you noticed everything; at 20 you cannot watch every source at once, so coverage gets thin exactly when there is more to cover.
• Check calls get missed. Tracking 20 trucks means 20 separate "where are you, are you on time" loops a day. Miss a few and you find out about a late delivery from an angry broker instead of from your own driver. The customer-facing damage is out of proportion to how small the miss felt.
• Deadhead creeps up. When you are reacting load by load instead of planning the next move for each truck, empty miles accumulate quietly. Industry deadhead runs somewhere in the 15-30% range, and at a truck cost of about $2.26 a mile (ATRI, 2025 data for 2024) every empty mile is a direct hit. A few extra points of deadhead across 20 trucks is a number that shows up at year end.
• Nobody has the full picture. This is the one that hurts most. The owner asks "how are we doing this week" and the answer is a guess, because the real state of the fleet is scattered across one person's memory, a spreadsheet, a few text threads, and a pile of rate confirmations. You cannot manage what you cannot see, and at 20 trucks you have quietly stopped being able to see it.

Notice what these have in common. None of them is a judgment problem. The dispatcher still knows which loads are good and which brokers to trust. The problem is throughput and visibility: too many parallel things to watch, and no single surface that shows the truth. That is the kind of problem software is genuinely good at, and it is why the right first move is to automate the watching and the surfacing, not the deciding.

What to automate first

When you hit the breakpoint, the temptation is to buy a big TMS and digitize everything at once. Resist that. A 20-truck fleet that tries to automate its whole operation in one move usually ends up with a half-configured system nobody trusts. Automate in the order the pain actually bites. Here is the ordering that holds up.

First, load search. This is the highest-leverage thing to take off a human's plate, because it is pure watching and your dispatcher is the bottleneck. Instead of one person tabbing between boards and broker portals between phone calls, let software watch every source continuously and surface the loads that fit your lanes, your equipment, and your rate floor. The dispatcher stops hunting and starts choosing from a filtered list. Searching many sources from one place is what a tool like Load Hub is built for, and its alert layer, Load Radar, pushes the matches at you instead of making you go look. This alone closes the "good loads slipped by" gap, which is usually the most expensive of the four.

Second, status updates and check calls. The 20-times-a-day tracking loop is repetitive, time-sensitive, and easy to review, which makes it the textbook case for automation. Automated tracking and status pings free the dispatcher from manually chasing every truck and only surface the ones that are actually off-plan: running late, off-route, sitting too long at a dock. You go from polling all 20 trucks to handling the two or three that need attention. The detention bill alone (an estimated $1.1-1.3 billion a year across the industry) is a reminder that catching a stuck truck early is worth real money.

Third, the action queue. Once search and status are handled, what is left is the list of things that need a human: the counteroffer to send, the driver to assign, the broker waiting on a callback, the appointment to confirm. The win here is not deciding for the dispatcher; it is collecting every pending action into one prioritized queue so nothing falls through. This is the visibility fix. Instead of the fleet state living in someone's head, it lives on a screen the owner can also read. That is the difference between guessing how the week is going and knowing.

Do these three in order and you have addressed every one of the four breakdowns without hiring. What you have not done, deliberately, is hand over judgment. Search hands you a filtered list; you still pick. Status flags exceptions; you still decide what to do about them. The queue organizes the work; a human still works it.

Keeping the human in the loop

It is worth being plain about this, because the pitch for dispatch automation often oversells it. Software should not be accepting loads on its own, sending unreviewed commitments to brokers, or assigning drivers without a person signing off. At 20 trucks your relationships and your rates are the whole business, and a single bad automated commitment can cost you a broker you spent two years building trust with. The job of automation here is to remove the work that does not require judgment, so the human has more time for the work that does.

The reason this matters more than usual at your size is leverage. A 50-truck fleet with three dispatchers can absorb a mistake; the other two desks keep running. At 20 trucks with one overloaded dispatcher, a wrong automated counteroffer or a missed exception lands directly on your reputation, and you feel it. So the right design keeps every outbound commitment under human approval. The freight market has roughly 27,000 brokers and broker margins run around 13.5% (DAT, 2023); your edge in that negotiation is judgment and relationship, and neither is something you want to automate away. Hand the machine the watching and the drafting. Keep the deciding.

If and when you do add a dispatcher later, this same setup makes that hire far more productive, because they inherit a system that already shows the full picture instead of a spreadsheet they have to reverse-engineer. Automation first, then headcount, is almost always the cheaper path through the 20-truck wall, and it is the one that does not require betting on a new person before you have your own operation under control.

The takeaway

The 20-truck breakpoint is real and it is predictable: the manual approach that carried you from five trucks runs out of capacity, loads slip, check calls get missed, deadhead creeps, and the full picture disappears. The fix is not to hire reflexively and it is not to automate everything at once. It is to automate in order: load search first, then status and check calls, then the action queue, leaving every real decision with a human. That sequence closes the specific gaps that open at this size while keeping your judgment and your broker relationships exactly where they belong. If you want a single place to start, start with load search, because it is the most expensive gap and the easiest to close. Numeo's AI Hub is built to find, rank, and surface loads under the dispatcher's control, which is the right shape for a fleet at the breakpoint, not a black box, but a faster pair of hands.