Leadership wants numbers. Auditors want records. Budget cycles demand defensible forecasts. And fleet managers are expected to produce all of it — accurately, on demand, with data that tells a clear story.
Here’s the problem: a lot of that data is wrong.
Not because anyone was careless on purpose. Wrong because of years of entry errors, system migrations that didn’t go cleanly, and workarounds that made sense at the time but quietly degraded the foundation. And once your data’s off, every decision you make from it is off too.
If you’ve ever looked at a report and thought, “Something doesn’t look right here,” that instinct is worth trusting. Let’s talk about where it comes from and what to do about it.
The U.S. Air Force ran an MS-DOS-based fleet management information system until 2016. Then they made the jump directly to a web-based FMIS. Big transition. Hard lessons.
Chief Master Sergeant Adam Walker, who oversees vehicle management for the Air Force, put it plainly: “We learned the hard lessons of garbage in and garbage out. We didn’t really give great guidance as a community to what our team was doing. We did the best we could at the time, but it didn’t meet the mark.”
That was a decade ago. They’re still working through it.
Think about that for a second. The Air Force manages 100,000 vehicles across 175 locations around the world. They have the budget, the talent, and the institutional commitment to fix problems at scale. And data quality is still an active, ongoing priority, not something they checked off and moved on from.
If it’s that hard for them, it’s worth taking seriously in your operation too.
Poor data quality isn’t just a reporting headache. It shows up in real, operational ways.
Maintenance decisions get made on incomplete history. When a vehicle’s records are patchy or inaccurate, your tech can’t tell what’s been done, what’s overdue, or what patterns of failure might be developing. PM compliance rates become unreliable. Work gets duplicated or skipped. Problems that should have been caught early aren’t.
Replacement cases fall apart under scrutiny. Fleet managers are increasingly expected to back up replacement requests with data: cost-per-mile figures, utilization trends, maintenance history. When those numbers are shaky, the case doesn’t hold. You either lose the budget argument or you win it for the wrong vehicle.
Budget forecasts drift from reality. Procurement timelines, parts spend, labor costs — all of it depends on data that reflects what’s actually happening. Bad inputs produce budgets that miss, and not in small ways.
Marc Canton, VP of Fleet Strategy at RTA Fleet, described a real example that stuck with him. An analyst was running a lifecycle analysis using a customer’s data. The results kept coming out wrong. After a lot of back-and-forth, they finally found the problem: a labeling misunderstanding. The customer had entered “year one” as the most recent year. The analyst had read it as the oldest. Maintenance costs go up over time, not down, so the entire analysis was running in reverse. Nobody caught it until significant time had been wasted.
“Sometimes it’s worse,” Canton said. “It’s better to have no data at all than to have bad data leading you in the wrong way.”
That’s a hard thing to hear. But it’s true.
Fleet data problems tend to cluster around the same failure points, regardless of fleet size or sector.
System transitions without a real data plan. Every FMIS migration is a risk. Historical data migrates incorrectly, incompletely, or in a format that strips out important context. Without deliberate validation before and after the move, quality problems get baked into the new system from day one.
No clear standards for data entry. When your technicians, coordinators, and managers aren’t aligned on exactly how to enter data, what counts as a completed PM, how labor hours get logged, which fault codes apply where, the data they produce is inconsistent. And inconsistent data can’t be reliably analyzed.
Institutional knowledge that lives in people, not systems. Most fleet operations have a significant amount of context that never gets documented. It lives in the heads of experienced staff who know why certain vehicles behave a certain way, which vendors you can trust, which data fields have always been used inconsistently. When those people leave, that context disappears, and the data they left behind gets harder to interpret.
Workarounds that quietly obscure reality. When a system is hard to use, people find ways around it. That usually means data doesn’t get entered at all, or goes in the wrong field with the wrong category. Over time, the system drifts further and further from what’s actually happening in your shop.
Data quality problems and the deferred decisions they cause don’t just affect reporting. They create real operational pain.
Chief Master Sergeant Walker described what happens when replacement cycles get stretched because decision-makers didn’t have the data to make the case for action early enough. Air Force mechanics were sourcing parts from junkyards. OEM support had ended for vehicles that had been kept in service too long. The mission still had to run, so the teams in the shops found creative solutions, solutions nobody wanted to be relying on.
That story isn’t unique to military fleets. Public fleet managers across the country are running vehicles well past their optimal replacement windows because the data available to make the replacement case wasn’t compelling enough to win the budget conversation.
The right data, in the right format, at the right time, changes that conversation. The wrong data (or no data) leaves you sourcing from junkyards and hoping for the best.
Improving fleet data quality isn’t a project with an end date. It’s a discipline. The fleets that do it well share a few consistent practices.
They treat data entry as part of the job. In well-run shops, logging accurate work order data, completing inspection records, and recording parts usage correctly isn’t optional. It’s understood as a core part of the technician’s role — not something that gets done when there’s extra time.
They invest in usability, not just capability. A powerful system that’s difficult to use will produce poor data because people will avoid it. The best FMIS platforms make correct entry the path of least resistance, not the most effortful option. At RTA, that’s a design principle, not an afterthought. Workflows are built around how fleet teams actually work, so the right way to enter data is also the easiest way.
They audit data regularly, not just after something goes wrong. Scheduled reviews of PM completion records, work order accuracy, and parts logs catch problems before they compound. The Air Force described this as an active, ongoing priority, not a one-time cleanup.
They connect data to decisions visibly. When your team can see how the data they enter today shows up in the reports, the budget requests, and the decisions made six months from now, the motivation to enter it accurately increases. Data quality improves when people understand why it matters.
The Air Force is actively working to integrate AI into their vehicle management operations, using it to process data at scale and support predictive maintenance decisions across 100,000 vehicles.
Chief Master Sergeant Walker was candid about the prerequisite: the data feeding the AI has to be good. Garbage in, garbage out doesn’t become less true when AI is involved. In fact, it gets worse: an AI system trained on bad data will produce bad outputs, often with more confidence than a human analyst would bring to the same wrong conclusion.
If you’re thinking about AI applications in your fleet operation, treat data quality improvement as Phase 1. Not something to circle back to after the AI is already deployed.
Not all fleet management systems are built with the same priorities. Some are designed for commercial fleets and retrofitted for public sector operations. Some are powerful but difficult to use in the day-to-day flow of a real shop.
RTA Fleet360 is built specifically for public-sector, in-house fleet operations. The workflows are designed around how fleet teams actually work, so correct data entry is the natural path, not the one that requires extra steps and extra time. If you’re evaluating platforms, look beyond the feature list. Ask how each system handles data entry inside the workflow, how it surfaces quality issues proactively, and how it turns historical data into something you can actually use in a budget conversation.
Fleet data quality is a discipline problem, a training problem, and a culture problem. Technology can support it or undermine it. but it can’t create it on its own.
The fleets building operations they can actually defend, in budget meetings, in audits, in conversations with leadership, are treating data quality as a foundational practice, not a cleanup project they’ll get to eventually.
The data you have is only as good as the processes behind it. And the decisions you make are only as good as the data.
Start there.
This article was inspired by a recent episode of our podcast. Check out the full episode for even more tips and tricks: