A fleet dashboard with twenty tiles is a dashboard nobody reads. Most of what is on it is there because it was easy to compute, not because anyone acts on it. These are the six that survive the test of changing a decision, and the trap each one carries.
The test
Before a number earns a tile, ask: when this moves, what does somebody do differently?
“Total distance this month” fails. It goes up when work goes up. Nobody does anything about it. It belongs in a report, not on a dashboard.
“Litres per hundred kilometres, by vehicle, against last month” passes. When it moves, somebody looks at that vehicle.
1. Consumption per 100 km, by vehicle
Why: fuel is the largest controllable operating cost in most fleets, and per-distance normalisation is the only way to compare a van doing city rounds with one doing motorway runs — imperfectly, but usefully.
What it tells you: a vehicle drifting upward is a mechanical problem, a driver change, or a route change. All three are worth a question.
The trap: comparing across vehicle classes. A 40-tonne truck and a panel van do not belong in the same ranking. Group first, then compare within the group.
2. Idling as a share of engine hours
Why: it is the purest waste there is — fuel burned, engine hours accrued, nothing moved — and it responds faster to attention than almost anything else on this list.
What it tells you: high idling is either habit, or a process problem where drivers wait with the engine running because of a loading delay you could fix.
The trap: refrigerated units, PTO work and cab heating in winter are legitimate stationary running. Exclude them explicitly or you will be coaching drivers for doing their jobs, which ends the programme.
3. Utilisation: days used against days available
Why: the metric most likely to save real money, and the one least often on a dashboard. A vehicle used eleven days a month is a lease payment, insurance and depreciation for nothing.
What it tells you: whether you have too many vehicles. The answer is often yes, and nobody knows because nobody looks.
The trap: seasonality and spare capacity. A vehicle at 40% utilisation may be the one that covers breakdowns. Judge over a quarter, not a week, and know which assets are deliberately slack.
4. Fuel events that are not consumption
Why: a drain buried in a monthly consumption figure is invisible, which is exactly what makes it profitable for whoever is causing it.
What it tells you: a count, per vehicle, per month, of level drops that were too fast to be burn — the rule is a threshold on rate of change.
The trap: treating it as an accusation. It is a lead. Sensor faults, tank geometry and refuelling artefacts all produce false positives, and a person has to look before anyone is spoken to.
5. Harsh events per 100 km, by driver
Why: it correlates with wear, fuel and — loosely — risk, and it is the input to any coaching programme.
What it tells you: who to talk to. One conversation with the bottom four drivers beats a memo to forty.
The trap: three of them. Do not rank across incomparable routes; urban work generates more events than trunking by nature. Do not publish a league table, because people optimise the score rather than the driving. And do not let it drive an automatic consequence — behavioural data about a named person carries obligations, and an automated decision with significant effect brings Art. 22 into play.
6. Data completeness: vehicles reporting today
Why: every other number on this list is wrong if a tenth of the fleet stopped reporting last Tuesday. This is the metric that tells you whether to believe the others.
What it tells you: devices that have gone quiet — a failed tracker, a disconnected harness, a SIM out of credit, an ingest pipeline that stopped.
The trap: none, really. This is the one nobody puts on the dashboard and everybody should. The failure mode of telematics is silence, and silence looks exactly like a quiet week. We learned this one the hard way: our own ingest stopped for five hours before anybody noticed, and everything downstream looked plausible the entire time.
What to leave off
- Total distance, total fuel, total trips. Volume, not performance. Reports, not dashboards.
- Live vehicle count. It is on the map already.
- Anything with a target nobody set. A gauge with a green zone that came from the software’s defaults is decoration.
- Averages without a distribution. A fleet averaging 28 l/100km may be uniformly mediocre or may be three vehicles ruining an otherwise good number. Only the second is actionable, and the average cannot tell you which you have.
Make them comparable, then make them recurring
Two operational points matter more than the choice of metrics.
Group before you compare. By vehicle class, by route type, by depot. A ranking that mixes them is noise wearing a chart’s clothing.
Send it, do not host it. A dashboard someone must remember to open is a dashboard read in the first week and never again. A report that arrives by email on Monday morning is read. Routly’s reports are built, saved, scheduled and emailed for exactly this reason — the arriving is the feature.
Frequently asked
How many tiles should a dashboard have?
Six is plenty. If a seventh is genuinely needed, something else should come off.
What period should I compare against?
Previous month for consumption and idling, previous quarter for utilisation. Week-on-week is noise for most fleets.
Who should receive what?
Fleet manager gets all six monthly. Depot supervisors get their own depot weekly. Drivers get their own numbers and nobody else’s. Finance gets utilisation and total cost, quarterly.
Can I build these myself?
That is the point of a KPI board where you pick a metric, a grouping and a shape — so a new question does not require anyone to ship you a new report. The demo has one, on generated data, with no account required.
