The cost of a bad decision is the gap between what you did and the best thing you could have done instead. For three years in truck fleet management, that gap was narrow, as rates were lower across the board. Fleets could be graded on numbers that reward activity rather than value, and the error stayed small because the alternative was, too.
That gap is widening. According to DAT’s June 2026 Truckload Volume Index, the national average van spot rate hit $3.00 per mile, topping the contract rate of $2.89, the first time spot has led contract since February, 2022. A 500-unit fleet making thousands of assignment decisions a week against a metric that rewards activity makes the same bad decision thousands of times.
The numbers most fleets are graded on were built for the market that just ended. They include the following:
Operating ratio, and the decisions it can’t see. Operating ratio is operating expenses divided by operating revenue. An OR of 92 means the fleet spent 92 cents to earn a dollar. It reports a result and says nothing about how that result was produced.
A planner assigns a driver to a load. A pricing analyst commits a lane in a bid. Each is defensible on its own, but made by someone with a partial view of the network and a scorecard that rewards their own function. Thousands get averaged into one figure with the reasoning stripped out. Two fleets can post the same OR from opposite operations: one from sound assignments, the other from shedding freight or reducing headcount.
A load that made money isn’t the same as the best load that driver could have taken, and once it’s dispatched, the alternatives are gone. Doing that comparison for every driver across hundreds of trucks is more than a planning floor can manage, so it gets done by instinct, differently in every region and shift. Automating the comparison is what makes it consistent, and consistent decisions are what let the OR finally tell you something about how the fleet is being run.
Fleet size, and the drivers who make it real. Fleet size is the tractor count: how carriers are financed, how they’re ranked, and how capacity gets committed in a bid. But a tractor is only capacity when a qualified driver is sitting in it.
The driver pool is shrinking under English language proficiency enforcement, non-domiciled commercial drivers’ license (CDL) restrictions, and the closure of training schools. A 600-truck fleet with 510 seated drivers is a 510-truck operation. Hours of service, home time and location decide whether a seated driver is capacity this week. None appears in a tractor count.
Count what you can actually run: available driver hours, and the same number in front of the recruiter, planner and whoever answers the request for proposals. It moves constantly, because every dispatch decision takes hours and repositions drivers. A fleet that sees next week's real capacity before committing stops is making promises it can’t keep.
Driver turnover, and the capacity it quietly removes. Turnover is the annualized rate at which drivers leave. It sits with human resources and gets managed with pay and recruiting spend. It’s a capacity number handled as a personnel number.
A driver who leaves takes a seat out of service for weeks, and a rate that shifts a few points changes how much freight the network can carry. But a driver assigned poorly, sent home late, or handed the loads nobody wants doesn’t pay. Those decisions get made daily by planners measured on coverage, never on retention.
Measure turnover where it happens rather than fleet-wide, and price it as capacity lost rather than recruiting to fund. Home time and duty cycles are inputs to the assignment, not constraints to work around afterward. When they sit inside the decision, the fleet stops trading retention for coverage without seeing the trade.
Regional performance, and the network cost it conceals. Most enterprise fleets are measured by region, each with its own leader, P&L and scorecard. But regional boundaries are administrative, not economic. Freight doesn’t respect them, and neither does the equipment.
This is the silo problem in its purest form: the reporting structure itself. A region can hit every target it owns by holding trucks inside its own borders, declining to release an asset worth more somewhere else, and pricing freight to its own margin rather than the network's.
Measure the network-wide result alongside the regional ones, and specifically what local optimization costs everyone else. When assignments are evaluated against the whole network rather than a boundary, the conflict stops being a judgment call.
Loads per planner, and what the planner is optimizing for. This is how much freight a planning organization covers per head. Growing a book of business has meant growing the planning floor, and that math gets expensive. The issue is that throughput counts decisions without saying anything about them.
Weighing every available driver against every available load, and what each choice does to the network next week, isn’t something anyone can hold in their head. It gets approximated with rules of thumb that vary by planner, region and shift. Two planners handed the same load will cover it differently. Both are defensible, and neither is measured.
Stop asking planners to do the comparison manually. When the decision comes off their plate, their judgment moves to where it’s irreplaceable: the customer exception, the driver situation, and the constraint that never made it into any system. Throughput rises because the comparison is consistent, not because people are working faster.
Every one of these five numbers reports a decision after the fact, once it has been averaged with thousands of others made by people who each saw only their own piece of the network. The fleets that come through this cycle intact will be the ones that stopped grading these numbers at quarter end and started evaluating each decision against the whole network as it happens.
Erica Frank is senior vice president of marketing at Optimal Dynamics.
