Start with cost per mile
You can name your van payment and roughly your fuel spend without looking anything up. Cost per mile takes an afternoon, and it is the number your decisions actually need.
That gap is why fleet decisions feel like guesses. “Can I afford another vehicle” has no answer without it. Neither does “is this contract worth running,” because a rate per mile only means something next to the cost per mile underneath it.
The arithmetic is not hard. Take twelve months, one vehicle at a time:
(payment or depreciation + insurance + fuel + maintenance + tires + registration) ÷ miles driven
Do it per vehicle, not for the fleet. A fleet average hides the one van that is eating your year. And count maintenance as what the breakdown actually cost you, including the tow and the routes you covered some other way, not just the invoice from the shop.
What comes out is one number per van, in dollars and cents. Once you have it, three decisions that used to be arguments become arithmetic: which vehicle belongs on your long-mileage runs, which broker contract clears its own cost, and whether the next van is a purchase or a mistake.
Count what can run tomorrow
Ten vans covering ten routes is a fleet of ten with no room for a dead battery.
The number that matters is how many vehicles can run a route tomorrow morning, and it is always lower than the number on your insurance policy. Subtract whatever is in the shop, whatever is waiting on a part, and whatever has a lift your drivers have quietly stopped trusting.
Operators resist spares because an idle van looks like waste. Run the comparison honestly and it usually isn't. A spare costs you payment and insurance and very little else, because it burns no fuel and no tires while it sits. A missed route costs the trips on it, the on-time performance the broker is scoring you on, and sometimes the facility itself, which called another provider at 6am and stayed with them.
The right ratio depends on how old your fleet is and how hard you run it, so find yours by counting. Over the last six months, how many mornings started a driver short, and what did you do about it? If the answer is “called around and moved trips,” your spare ratio is already telling you what it should be.
Before you buy another van
Volume is up, dispatch is under water, and a broker is asking whether you can take more. The instinct is to buy capacity. Ask three questions first, in this order, because the first two are free.
Are the vans you already own full?
Trips per vehicle per day is the number that separates a fleet problem from a routing problem. Across the fleets we see, a vehicle runs about eight trips a day on a six-day week. If yours are running five, a seventh van fixes nothing. You get seven vehicles running five trips a day: the same revenue spread over another payment and another insurance line, and a dispatcher with one more thing to keep straight.
Check that gap before anything else. Closing it costs you better planning. Adding a van costs you every month for five years.
Is it capacity you are short of, or the right kind?
Capacity is measured by trip type, and the types rarely run out together. The common version: refusing wheelchair trips in the morning while ambulatory vans sit in the lot. That reads as needing more vehicles when it is really needing a different mix.
Pull a quarter of completed trips, split them by type, and put that split next to your fleet. If your trip mix has drifted toward wheelchair or stretcher work and your fleet has not, you have your answer about what the next vehicle should be, and about which trips you should stop turning down.
Can the new van carry its own cost?
This is where your register earns its keep. A vehicle has to clear its cost per mile at the utilization you can honestly give it, not the utilization you are hoping for. Run the number on the trips that would go on it in month one, not on the contract you have been promised.
A van bought against a contract that lands two quarters late is a payment, an insurance line, and a depreciation schedule, all running against trips that do not exist yet.
Utilization and mix take an afternoon to check. The van is a five-year commitment. It should be the last question you ask, not the first.
Maintenance is a scheduling decision
Preventive maintenance is not really a maintenance problem. Your shop already knows the intervals. It is a scheduling problem, and the operators who get it wrong are usually not the ones skipping service. They are the ones servicing whenever a van happens to be free.
A vehicle in the shop on Tuesday morning and the same vehicle in the shop on Saturday afternoon are two different costs. One pulls a van out of your heaviest block and forces reassignments across three drivers. The other costs almost nothing, because your volume is already low.
So build the PM calendar against your own volume instead of the shop's convenience. Pull twelve months of completed trips, find your lightest weekday and the lightest block inside it, and put standing appointments there, booked a quarter ahead. A van with a scheduled slot in a quiet block is a rounding error. The same van breaking down at 7:40 on your heaviest morning is four late dialysis pickups and a call from the broker.
Credentials belong on that same calendar: vehicle inspections, registrations, and insurance renewals, each with enough lead time that nothing ever parks a van you were counting on.
The cadence
| What happens | How often | Who owns it |
|---|---|---|
| Vehicle register refreshed, cost per mile recalculated | Quarterly | Owner |
| Runnable-tomorrow count and short mornings reviewed | Monthly | Dispatcher |
| Utilization and trip mix checked against the fleet | Before any new vehicle | Owner |
| Service appointments booked into the light block | Quarterly, a quarter ahead | Dispatcher |
| Credential, inspection and registration sweep | Monthly | Whoever runs the office |
| Replacement triggers checked against the register | Quarterly | Owner |
| Re-keying audit on one trip, end to end | Once, then on any software change | Owner |
The replacement trigger
The most expensive vehicle in many NEMT fleets is the one the owner is proudest of: the old van with no payment on it.
Paid off feels free. What it actually does is trade a fixed, predictable cost for a variable, unpredictable one, at exactly the point in a vehicle's life when the variable cost starts climbing and the unpredictability starts landing on your dispatcher at 7am.
There is a clean trigger for this, and it takes one spreadsheet column. Track trailing six-month repair cost per mile on the old van. Compare it to the carrying cost per mile of the vehicle that would replace it: payment plus insurance, divided by the miles that route actually runs. When the first number crosses the second, the old van became the expensive one, and it did so a few months before you noticed.
| Track | Where it comes from | What it tells you |
|---|---|---|
| Repair cost per mile (trailing 6 months) | Shop invoices divided by miles over the same period | What the old van costs you to keep running |
| Carrying cost per mile on a replacement | Quoted payment plus insurance, divided by the miles that route runs | What the new van would cost in the same slot |
| Unplanned down days | Your own dispatch record: mornings that started a driver short | The cost the invoices never show |
| Odometer against useful life | Monthly reading per vehicle | How many months of warning you actually have |
Two things the arithmetic misses, and both push the same way. Repair cost says nothing about the routes you lost while the van was down. And a vehicle that fails unpredictably eats dispatcher attention you are already short of, which never appears on an invoice and is often the larger number.
Run the re-keying audit
Everything above is the half of this component you can touch. The other half is the software, and the bar for it comes down to three questions.
Three questions. Does it plan the day faster than your dispatcher can by hand? Does it keep up when the day changes, or does the plan die at the first cancellation? Does it hand billing what it needs without anyone re-keying it?
That last one is measurable, and it is the audit worth running this week. Follow a single trip from the moment it arrives to the moment the claim goes out, and count every time a person types information the system already had. The broker file lands and someone keys it into dispatch. The driver finishes and someone keys the completion into billing. Mileage gets entered a third time on the claim.
Count them honestly. The number is almost always higher than the operator guessed, because every hop was added one at a time and none of them felt like a decision. Each one is a place an error enters, and errors in NEMT do not surface as errors. They surface as denials, six weeks later, with half the resubmission window already gone.