
Preventive maintenance fails less because people don’t care, and more because the system can’t reliably answer simple questions: Which asset is this? How is it being used? When was it actually serviced? What did we learn?
When those answers are fuzzy, PM turns into maintenance theater: lots of scheduled activity, still plenty of surprises. The uncomfortable truth is that PM is mostly an information pipeline disguised as a wrench-and-checklist process.
A real-world pattern: the day the dashboard lied
Here’s a scenario that shows up in more fleets than anyone likes to admit.
A maintenance lead opens the CMMS on Monday morning and sees a comforting wall of green. PM compliance is “high.” The week is planned. The team is busy. Everything looks under control.
Then a machine goes down on a jobsite. Not a dramatic failure—just a quiet, expensive stop that triggers downtime, rescheduling, and a parts scramble. The team pulls the maintenance history to see what was missed… and the story doesn’t line up.
The asset exists twice in the CMMS. One record has the PM plan and a tidy trail of “completed” inspections. The other record holds the breakdown work order and parts consumption. Both records are “real,” but neither is complete. The dashboard was green because it was looking at the wrong version of reality.
This is why duplicate or fragmented asset records are so damaging: they don’t just create admin mess. They split the past, and when you split the past, you sabotage every decision that depends on it.
When teams stop trusting the data, they stop using the system. And when that happens, PM becomes calendar-based guesswork again.
PM is a chain of information decisions
Preventive maintenance looks deceptively simple on paper: “Service every X hours.” In real operations, it’s a chain of small decisions made under time pressure.
First, you need identity: one asset equals one truth. Then you need a meter you can trust: engine hours or mileage that arrive in time to act, and are consistent enough that people believe them. Then you need rules that match reality: OEM schedules plus clearly justified overrides for duty cycle, environment, and warranty constraints. Then you need signals you can interpret: inspections, fault codes, oil analysis, operator feedback. Finally, you need the closed loop: when work is done, the outcome has to feed back into the system so next month’s decisions get smarter.
Break any link, and “PM compliance” becomes a vanity metric. You can have a full schedule and still be unprepared, because the schedule is built on shaky information.
Why “more data” can make things worse
When PM isn’t working, the reflex is usually to add more: more sensors, more alerts, more dashboards. Sometimes that helps. Often it just produces higher-resolution confusion.
Maintenance information tends to live in pieces: CMMS/EAM, telematics, OEM or dealer portals, spreadsheets, invoices, and people’s heads. If those fragments can’t be reconciled into a consistent history, analytics projects spend most of their time on data preparation before they can deliver meaningful operational insight. Research in maintenance/repair/overhaul case studies repeatedly highlights the burden of fragmented datasets and data preparation work.
The “minimum viable record” that makes PM learn
One of the simplest improvements is also the least glamorous: define the minimum information that must be captured so maintenance history becomes useful, not just “closed.”
When work orders close without a meter reading, symptoms, cause, action taken, or meaningful notes, the system isn’t learning. It’s accounting. Over time, that creates the illusion of control—until something fails and nobody can confidently explain why, or how to prevent the next one.
That’s why CMMS data quality problems (like duplicates and missing fields) are not IT issues. They are reliability issues.
What “good” looks like (and why it’s boring)
High-performing PM operations usually have a few boring, repeatable mechanics:
There’s a single asset register that people actually trust. Meter data is fresh enough to drive scheduling decisions. Service rules are explicit and explainable. Work orders capture enough truth that the next person can understand what happened. Exceptions are reviewed deliberately, not reactively. And the loop closes: repeated failures trigger rule changes, inspection updates, or data fixes—not just more firefighting.
Why this matters: downtime is economically violent
Unplanned downtime isn’t just inconvenient. It’s expensive at a scale that turns “data quality” into strategy. Siemens’ “True Cost of Downtime 2024” report estimates the world’s 500 biggest companies lose almost $1.4 trillion annually through unplanned downtime, equivalent to 11% of revenues (their estimate).
Your fleet won’t match that number directly (industry and asset types vary wildly), but the direction is consistent: the cost of “information friction” gets paid in downtime, overtime, expediting, and missed utilization.
Closing thought
If preventive maintenance feels like a constant scramble despite “high compliance,” treat that as a diagnostic signal. The issue may not be effort or discipline. It may be that the information chain can’t support the decisions you’re asking it to make.
Sources
- Siemens (Senseye), The True Cost of Downtime 2024 (PDF): https://assets.new.siemens.com/siemens/assets/api/uuid%3A1b43afb5-2d07-47f7-9eb7-893fe7d0bc59/TCOD-2024_original.pdf
- FTMaintenance, CMMS Data Quality: How Bad Data Impacts Maintenance: https://ftmaintenance.com/cmms/cmms-data-quality/
- Pelt et al. (2019), Data analytics case studies in the maintenance, repair and overhaul industry (MATEC Web of Conferences PDF): https://www.matec-conferences.org/articles/matecconf/pdf/2019/53/matecconf_easn2019_04005.pdf