Every fleet platform ships with dashboards. Most are abandoned within a quarter, because they were designed around what the data warehouse could produce rather than around a decision somebody has to make.
Design by audience, not by data
Four audiences, four completely different needs.
Drivers and technicians need today: what is assigned to me, what is due, what is blocked. Not trends, not averages, not comparisons.
Supervisors and depot managers need exceptions: what is off-plan and needs intervention this week.
Fleet and operations managers need trends and structure: is performance improving, where is the cost concentrated, what should we change next quarter.
Finance and executives need outcomes and comparisons: cost per unit, budget variance, and whether the investment paid off.
A single dashboard serving all four serves none of them. Build four, each with a stated owner and a stated decision it supports.
For every tile on a dashboard, name the decision it informs and the person who makes it. Tiles that fail this test are why dashboards get abandoned — they turn a management tool into wallpaper.
What belongs on each
Depot manager — daily exception view
- Vehicles off road now, with reason and expected return
- PM services due in 7 days that are not yet booked
- Open defects over 48 hours old, by severity
- Documents expiring in 30 days
- Yesterday's routes that ran over plan by more than 30 minutes
Everything on this list is actionable today. Nothing on it is a trend.
Fleet manager — weekly management view
- Availability percentage, trended, by depot
- PM compliance (services completed within tolerance)
- Cost per mile, trended, with the top five outlier vehicles named
- Unplanned versus planned maintenance ratio
- Repeat repairs within 30 days
- Overdue compliance items
Finance — monthly
- Total cost of operation against budget, by category
- Cost per mile and per operating hour, by vehicle class
- Depreciation and residual position
- Vehicles beyond optimal replacement point
- Warranty recovery achieved versus available
Executive — quarterly
Five numbers at most: cost per unit of output, availability, safety incidents per million miles, compliance exposure, and the trend on each. Anything more gets skimmed.
The metrics that matter most
| Metric | Definition trap |
|---|---|
| Availability | Against planned operating time, not calendar time |
| PM compliance | Within tolerance window, not merely "eventually done" |
| Unplanned maintenance ratio | Unplanned cost ÷ total maintenance cost; the clearest programme health signal |
| Cost per mile | State whether it includes depreciation, fuel and overhead — three fleets, three definitions |
| Mean time to repair | From defect report to return to service, not from work order open |
| Repeat repair rate | Same system, same vehicle, within 30 days; the best quality indicator you have |
| Downtime cost | Requires a per-vehicle-per-day revenue or replacement figure; agree it once with finance |
Write the definitions in a data dictionary published alongside the dashboards. Undefined metrics generate meetings about whether the number is right instead of what to do about it.
Why dashboards fail
No owner. A dashboard without a named owner degrades as soon as the underlying data model changes.
Mixed audiences. An executive tile next to a technician tile means neither audience trusts the page.
Vanity metrics. Total miles driven, total work orders closed. Impressive, unactionable.
Averages that hide distributions. A fleet averaging 92% availability with two vehicles at 40% has a specific problem, invisible in the average. Show distributions and outliers, not just means.
No trend context. A single number without its history cannot tell you whether to act.
Refresh mismatch. A daily-refreshed cost tile invites people to react to noise; a monthly-refreshed exception tile is useless.
Building the reporting layer
Three practical rules:
- Get data out of the platform. Insist on API access or a data export capability. Fleet platforms are good at operational reporting and rarely good at analysis across systems. Combining fleet, finance and route data usually requires your own layer.
- Model once, present many times. Define cost per mile in one place. Two teams computing it separately will produce two answers and spend a year reconciling them.
- Automate delivery. A report emailed every Monday at 07:00 gets read; a report someone must log in to find does not.
Common questions
Should we use the vendor's reporting or our own BI tool?
Use the vendor's for operational, exception-based reporting where the data lives. Use your own BI tool when combining fleet data with finance, HR or route data, or when you need history beyond the platform's retention. Most mature fleets end up with both.
How often should dashboards be reviewed?
Review the dashboards themselves twice a year: which tiles were used, which decisions changed, what is missing. Dashboards are products with a lifecycle, not artefacts that are finished at launch.
What is the single most useful fleet report?
The unplanned-versus-planned maintenance ratio, trended over 12 months. It captures the health of the whole maintenance programme in one line, and it moves slowly enough that a change means something real.
How do we stop people arguing about the numbers?
A published data dictionary, one owner per metric, and one source of truth. Most disputes are definition disputes, and they are cheap to prevent and expensive to relitigate every month.
Should drivers see fleet dashboards?
Drivers should see their own data — their inspections, their assignments, their vehicle's status. Comparative dashboards ranking drivers require careful design and a coaching culture, or they become a source of grievance rather than improvement.
