EMI3News & insightInsight
The missing link: sensory data and sensor data
Sites invest heavily in oil sampling, vibration analysis and sensor technology, then lose the observations of the people standing next to the machine. The two belong in one record.
Maintenance has moved a long way. Reactive repair gave way to planned preventive servicing. Planned servicing was joined by condition monitoring: oil sampling, vibration analysis, thermography, sensors reporting continuously on asset health.
And yet teams still describe the same two frustrations: defects that were noticed but never recorded, and an incomplete picture of what condition an asset is actually in.
Where the information goes
Paper service sheets and defect notes get lost in filing cabinets, or never make it into the CMMS at all. A technician who spots something at the end of a long shift, with no time and no tool to hand, does not raise it. The observation was made. It simply was not captured.
The gap is not between people and machines. It is between sensor-based data and sensory-based data.
What a person detects that a sensor does not
Instrumentation is excellent at what it measures and blind to everything else. The technician standing at the machine is running four channels no sensor covers:
- Sound: an unusual noise that says a bearing is wearing.
- Touch: a subtle vibration through a handrail pointing to misalignment.
- Smell: an odour signalling overheating or contamination.
- Sight: the weep, the witness mark, the fretting that has not yet become a failure.
These are real measurements taken by a competent instrument. They are simply not written down anywhere a system can use them.
Closing it
Three things have to be true. Capture has to happen in the field, on a device the technician already has, prompted by the task rather than left to memory. The finding has to flow straight into the record rather than being re-entered later. And the two data sets have to sit side by side, so a sensory observation can be read against the sensor trend for the same asset.
Do that and the payoff is not more data. It is hidden defects identified before they escalate, decisions made against a complete picture of asset health, and less downtime bought with work that was already being performed.
Was this useful?
We read every response. It decides what we write next.
We will not add you to a mailing list.
Keep reading
Product
Inside the limits, and still going wrong
A condition monitoring question captures a number and a photograph. The platform trends it across the fleet, watches the rate of change, and raises an alert while the reading is still comfortably inside its limits.
4 September 2026 · 7 min read
Insight
Your CMMS is not the execution layer
Enterprise systems hold the asset record. Planning systems schedule the work. Neither of them sees the work being performed, and that gap is not a reporting problem to be fixed with another dashboard.
12 August 2026 · 8 min read
Insight
Chaos, constrained: why maintenance behaves like a complex system
Small deviations at the start of a shift produce very large differences by the end of it. You cannot remove that sensitivity, but you can bound it, the way aviation and surgical systems do.
29 July 2026 · 6 min read