Maintenance Cost per Vehicle and the Three Things It Conflates

Why this matters

Maintenance cost per vehicle is the number most shops reach for when somebody asks whether the fleet is getting expensive, and as an answer it is close to useless. It blends three causes that call for three different responses: how old the fleet is, how hard it is being used, and how it is being driven and loaded. An eight-year-old truck covering real distance and a two-year-old van being flogged produce the same shape of number and need nothing in common from you.

In the fleet worked below, the raw figure points squarely at the oldest truck, which is the least surprising finding available and the one you can do least about, and never notices the two-year-old van spending two and a half times what its identical twin spends. That is the normal outcome of averaging three causes, not a freak case.

What the figure is made of

The numerator is every maintenance line recorded in the window. The denominator is the number of vehicles, and that is where shops differ without realising it. Two bases are defensible and they answer different questions:

• Vehicle-years of ownership. Each vehicle counts for the fraction of the window you owned it. Make this the default: it survives a sale and a purchase mid-window, and it answers "what did it cost to keep a truck on the road last year."
• The fleet you will still own. Restrict numerator AND denominator to the vehicles staying. This answers "what should I budget next year," and it is only correct when both sides are restricted together.

What almost every shop actually computes is neither: the whole spend log over today's active vehicle count. Throughout, index maintenance spend in units where 1.0 unit is what one standard oil-and-filter service costs on this fleet's light van, so the figures travel.

The numerator and the denominator count different fleets

The spend log is a record of work done. The active-vehicle count is a snapshot of what you own the day you pull the report. Sell a truck in month four and its repairs stay in the log for the rest of the year while it leaves the divisor, so the figure rises with nothing having changed.

Five active vehicles, twelve-month window:

Truck F was sold in month four and carries 9.0 units of spend before the sale. The log total is 46.0 plus 9.0, or 55.0 units.

• What gets reported: 55.0 over 5 active vehicles is 11.0 units per vehicle.
• Vehicle-years: F was owned 4 of 12 months, or 0.33 vehicle-years, so the divisor is 5.33. 55.0 over 5.33 is 10.3 units.
• The fleet you keep: strike F from both sides. 46.0 over 5 is 9.2 units.

The reported figure is 11.0 against a budgeting figure of 9.2, or close to 20 percent high, and 11.0 against the vehicle-years figure of 10.3, or about 7 percent high. Neither correction is large on its own. Both run in the same direction every single year, which is why a fleet that turns vehicles over looks like it is losing a battle it is not fighting.

The mirror case flatters you. Add Van G in month ten, spending 1.0 unit in its three months: the log is 47.0 over 6 active vehicles, so the figure falls to 7.8 units, down about 15 percent from the 9.2 the five established vehicles actually cost, with nothing improved. On vehicle-years the divisor is 5.25 and the figure is 8.95, which is the honest read. A basis that moves when you buy or sell is reporting your purchasing, not your maintenance.

Three causes averaged into one number

Once the population is clean, the figure still carries three independent drivers, and they belong to three different people in the shop. Age goes in a replacement plan and nothing you do this quarter touches it. Usage goes to dispatch, and more miles meaning more spend is the system working. Driving and loading is the only one of the three that a conversation next week can change.

An average over all three is unactionable because you cannot tell which one moved. The fix is not a better average, it is two cuts that remove one driver at a time.

Cut one: divide by usage, and the age curve appears

Divide each vehicle's maintenance spend by its own distance, and usage stops competing with age. This is the maintenance slice only, not the all-in cost per mile that includes fuel, insurance, depreciation and financing, which a sibling article owns and which is the figure to use for replacement timing.

Ignore B for a moment and the age curve is clean: 0.33 at two years, 0.63 at five, 1.00 at seven and 1.42 at eight, so maintenance cost per mile roughly doubles by year five, triples by year seven, and passes four times the two-year figure by year eight. That curve is the useful part of the age driver, and on a fleet this size it rests on one vehicle per age point with the one contradicting vehicle set aside, so read it as a shape to confirm against two or three more years of your own records rather than as a replacement budget you can sign.

Take the fleet rate as total over total, 46.0 units over 63,000 miles, or 0.73 units per 1,000 miles. Averaging the five per-vehicle rates instead gives 0.84, because the highest rates sit on the lowest-mileage vehicles and an unweighted mean gives them the same vote as a van doing three times the distance. Use total over total for the fleet figure. See related below for why the two differ and what the size of the gap tells you.

Not every mile is the same mile

Dividing by distance removes volume, not duty cycle. Two caveats decide whether the per-mile number can be compared across vehicles at all:

• Calendar-driven items do not scale with miles. An annual service, an inspection, a battery and tyres replaced on age rather than tread arrive whether the van moves or not, so dividing them by a small distance inflates the rate. Truck E carries 6,000 miles and a full year of calendar items, so a meaningful share of its 1.42 is that arithmetic rather than a sick vehicle. Compare a low-mileage vehicle against its own prior years, not against a high-mileage sibling, or strip the calendar items out of both numerators first. Even corrected, E costs well above the 0.73 fleet rate per mile while sitting below the fleet average per vehicle, and what that points to is whether a spare covering 6,000 miles a year earns its place, which is a fleet-sizing question rather than a maintenance one.
• Stops per mile matter more than miles. A van doing dense residential work brakes, idles and cold-starts many times per mile; a van running long suburban legs does not. Where a vehicle idles to run on-board equipment, engine hours beat miles as the denominator, and most vehicles built in the last decade report them.

Cut two: hold age constant, and the four readings side by side

Vans A and B are the same model, bought together, two years old. Same age, same calendar exposure, same warranty position. Whatever separates them is not age, and once you divide by distance it is not volume either. What is left is how they are driven and loaded.

Read the rows in order and the finding tightens rather than collapses. On raw spend B is only 1.25 times A, which is why nobody noticed: B spent a quarter more while driving half the distance. Per 1,000 miles it is 2.5 times. Stripping the calendar items both vans carry equally makes it 2.7 times, so the correction that could have rescued B makes it worse, which is the direction that settles the question.

The last row is a deliberate stress test and it is one-sided by design. B's 7.5 units are a brake job at 3.0, tyres at 2.5, a routine service at 0.8 and suspension bushes at 1.2. Drop the single largest event and B still sits at 0.50 against A's uncorrected 0.33, or 1.5 times. Nothing is taken off A, so the row understates the gap by construction, and understating is what a flag should do. Report the 1.5x as the finding and retire the 2.5x headline: one brake job was carrying most of it.

What survives is brakes, tyres and suspension on a low-mileage van, and that combination has a short list of causes. Here is what each pattern points at and how to tell them apart:

B's pattern sits in the first two rows, which is a driver and route conversation and possibly a load audit. Before that conversation, check one thing: that B's miles are being captured. A van whose odometer is read once a year from memory can produce this exact signature with nothing wrong at all.

What this figure may authorise, and what it may not

A window shorter than twelve months will not carry any of this, because maintenance is lumpy and seasonal: one clutch or one set of tyres inside a quarter swamps everything else. Read the full year, then work down this ladder.

The portable rule underneath the ladder: never act on a maintenance figure whose denominator you did not choose. Vehicle count was chosen for you by whatever the records happened to hold on the day. Distance and age are choices, and they are the ones that can name a vehicle.

References

• See related: Tracking Cost per Mile Across a Mixed Fleet, for all-in cost per mile and reading the replacement trend
• See related: Maintenance Frequency Against Maintenance Cost, for what the event count adds to the spend total
• See related: Gross Margin Percent Is an Average of Averages, for why an unweighted mean of per-vehicle rates sits above the total-over-total rate
• See related: Right-Sizing Your Fleet to Actual Job Volume, for the low-mileage spare question
• See related: Assigning Vehicles to Drivers Fairly and Sensibly, for handling the driver conversation cut two produces