Guide

How to run a fuel program at a trucking company.

This is the operator's version: what to do, in what order, whether or not you ever work with us. It assumes you have a fuel card with negotiated pricing and nobody assigned full-time to making sure it gets used.

1. Baseline what you spend before you change anything

Pull twelve months of fuel-card transactions. For each one you want the date, the truck, the driver, the station, the gallons and the price paid. Ask your provider for it as a CSV; if all you can get is a PDF, keep asking, because a PDF is not a baseline.

Split those transactions into in-network and off-network using your provider's site list. The in-network share of that split is your compliance rate, and it is the single number the whole program moves. Work it out before you start, or you will have no way to tell later whether anything improved.

Do not skip this because you think you already know the answer. Until it is measured, the compliance rate dispatch has in mind is an estimate, and an estimate can be wrong in either direction.

2. Work out what a gallon actually costs you

The number on the sign is not the number you pay, and the gap between them is where a fuel program lives. Until you can state your net price at a given station, you cannot rank two stations against each other, and every later step in this guide is guesswork dressed up as a decision.

Start with the shape of your agreement, because there are two and they behave differently. One is a deduction off the posted retail price: you pay what the sign says, less an agreed amount. The other prices from a published benchmark plus the costs of getting fuel to that site, and ignores the sign entirely. The first moves with the station's own pricing, so a site that posts high can hand back a worse net price than a site that posts low with a smaller deduction. The second does not, which makes it comparable across a lane but harder to sanity-check at the pump.

Then add what the agreement does not cover. Transaction fees, out-of-network surcharges and any per-gallon service charge are part of what leaves your account, so they are part of the net price whether or not the pricing schedule mentions them. A stop that wins on the discount and loses on the fees is a stop you took for the wrong reason.

The consequence is the point of the whole exercise: the station with the biggest discount is not reliably the cheapest place for that truck to fill. Rank by net price, lane by lane, and the ranking will not match the one in your head.

3. Map your network onto your actual lanes

A discount network is only worth what your trucks can reach. Take your ten busiest lanes and mark where the in-network stations sit along each one. You are looking for the gaps: stretches where a truck running low has no in-network option within a reasonable detour.

Those gaps are where your off-network fills concentrate, and they are not the driver's fault. Some of them you can fix by talking to your provider about adding a site; some you fix by planning the fill earlier, on the side of the gap where you still have a choice.

4. Set a compliance target, and make it somebody's job

Pick a number and name an owner. Without a named owner, nothing stops the decision drifting back to the driver at the exit, and the drift is quiet, because nothing visibly breaks.

The target should be reachable from your baseline, not aspirational. If you are starting from a middling compliance rate, moving it steadily is worth more than announcing a number nobody believes.

5. Get the decision to the driver before the tank forces it

This is the part that decides whether the program works. Everything up to here is analysis; this is operations.

The stop has to reach the driver while they still have enough range to take it. That means the trigger is the fuel level and the lane ahead, not a morning plan — a plan made at 6am is wrong by lunchtime, because the truck moved and so did the load.

Use the channel the driver already reads. Whatever your drivers use to talk to dispatch is the right channel; a new app that a driver has to remember to open is not, however good the routing behind it is. Keep the message short enough to act on at a wheel: the station, the exit, the reason, and a way to say no.

Give them a way to decline. A driver who cannot decline will simply ignore the message, and you lose the signal that would have told you why.

6. Reconcile every month against the card statement

Match each fuel-card transaction back to the stop it was supposed to be. Three things fall out of that match, and all three are useful.

Transactions that match an assigned stop are the program working. Transactions with no assignment are the fills nobody planned, which is your remaining leak. Assignments with no transaction are stops that were sent and not taken, which is either a coaching conversation or a legitimate reroute you did not know about — and the match alone does not tell you which, so ask before you treat it as either.

Keep estimated and actual savings apart. An estimate made when the stop was assigned and a figure derived from the settled transaction are different kinds of number, and blending them produces a total that nobody can defend in a review — which means, in practice, a total that gets quietly ignored.

7. Coach on the numbers, not on the average

Fleet-wide compliance is a reporting number, not a management one. Break it down by driver, truck and lane to see whether the off-network volume is spread across the fleet or concentrated in a few drivers and lanes; only the breakdown tells you where to act.

Then ask why, per driver, before assuming it is discipline. Common and fixable answers: the driver runs a lane with a real network gap, the driver plans fills around a familiar stop that happens to be off-network, or nobody ever told the driver the program existed.

The three numbers worth tracking

Everything else is detail on top of these.

  • Compliance rate

    In-network fills as a share of all fills. The number the program moves, tracked the same way every month so the trend is real.

  • Recovery per truck per month

    Derived from matched transactions, labelled actual. Per truck rather than fleet-wide, so growth in the fleet does not disguise a flat program.

  • Unmatched transactions

    Fills with no corresponding assignment. This is the honest measure of what the program still is not seeing, and it should shrink.

Four mistakes worth avoiding

Starting with a dashboard. A dashboard shows a dispatcher what already happened; the program needs the decision to reach the driver before it happens. Reporting is the last part to build, not the first.

Rolling out to the whole fleet at once. Ten trucks for a month shows you what breaks, whether that is the data plumbing or the routing, while it is still cheap to fix.

Measuring against a retail price feed you do not own. Savings computed against a published average are an argument; savings computed as your own contracted discount per gallon on confirmed in-network fills are arithmetic from your own contract.

Letting the program run without an owner. See step four: without one, the program can fade without anyone noticing.

Questions

How far back should the baseline go?
Twelve months. A shorter window can be dominated by one bad quarter or one seasonal lane, and you will spend the following year arguing about whether the starting point was fair rather than about what changed.
Should compliance be measured in gallons or in transactions?
Pick one and say which, because the two answer different questions and diverge in a way that matters. Counting transactions tells you how often the program is followed; counting gallons tells you how much fuel it actually governs. A handful of large off-network fills barely moves the transaction count and moves the gallon share a lot more, which is why a fleet can look compliant by one measure and not the other.
What do we do about a lane with a genuine network gap?
Treat it as a planning problem rather than a compliance problem. A stretch with no in-network option inside a reasonable detour is not the driver's fault, and counting those fills against them teaches the fleet that the number is unfair. Either get the site added, or plan the fill before the gap while the truck still has a choice.
A stop was sent and the driver did not take it. Is that non-compliance?
Not on its own. It is an assignment with no matching transaction, which is either something to coach or a reroute the office did not know about. The reconciliation shows you the gap; only asking tells you which of the two it was.
Do we need software to start?
No. Every step above can be done with a card export, a spreadsheet and somebody whose job it is. Software and services change how much of it is manual and how fast the decision reaches the driver; they do not change what the work is.

Or have somebody run it for you.

DispatchFuel.ai does the work above as a service for trucking carriers. The pilot runs it on part of your fleet, with your data, so you can judge it on your own numbers.