Handling equipment generates useful data with almost none of the sensitivity attached to tracking people, provided the boundary is kept.
What it produces
Hours powered and hours in motion, which are different and both matter.
Distance and location, at least by zone.
Impact events, with severity and location.
Speed, and speed in defined zones.
Battery state and charging behaviour.
Fault codes and service intervals.
Access control: which licensed operator unlocked the machine, and whether the pre-use check was completed.
The operational uses
Fleet sizing from utilisation against demand by hour.
Parking and staging positions, so machines start near the work.
Maintenance by actual usage rather than by calendar, which is more accurate and usually cheaper.
Downtime quantified, which supports the spares argument.
Battery management, shifting charging out of peak demand.
The safety uses
Impact events as an early indicator of racking damage, layout problems and training needs.
Speed monitoring in pedestrian zones, which is a legitimate and defensible control.
Pre-use check compliance, which is a legal requirement in many jurisdictions.
Operator licensing enforcement, so an uncertified person cannot start a machine.
Proximity warning, where the system supports it, which is the strongest safety application in the whole field.
The boundary
Access control means the machine knows who is operating it, which makes every subsequent data point personal data about that person.
This is unavoidable and it is the point where care is needed.
Use it for what it is justified by: licensing enforcement, safety, and incident investigation.
Do not use it for productivity ranking of operators, which is the drift that turns a well-justified safety system into surveillance.
State the purpose limitation in writing and enforce it, because this is exactly where scope creep happens.
Impact data specifically
The most valuable and most misused signal.
Impacts indicate a problem, which may be layout, congestion, visibility, machine condition, training or pressure to work fast.
Using impact data primarily for individual discipline produces under-reporting — operators avoid triggering sensors, disable them, or do not report incidents.
Which destroys the safety value entirely, since the purpose is to find the conditions causing impacts.
Investigate the location and the circumstances first, every time. A cluster of impacts at one corner is a layout finding, not a personnel finding.
Tell operators that is the policy and hold to it, or the data degrades within months.
Practical deployment
Retrofit units exist for older fleets and are usually worth it.
Check what the fleet already produces. Many modern machines log more than the operation collects.
Integrate with maintenance, or the data produces reports nobody acts on.
Integrate with the task system, so utilisation can be compared against demand rather than reported in isolation.
Agree the data retention and access before deployment, since it contains operator-identified records.
Reading an impact cluster
The investigation order that determines whether impact data stays useful.
Location first. A cluster at one corner is a layout, visibility or surface finding.
Time second. A cluster at shift change is a scheduling finding.
Task third. A cluster during one activity is a method finding.
Machine fourth, since a machine with disproportionate impacts may have a steering, brake or visibility fault.
Operator last, and as a training question rather than a disciplinary one, which is what keeps the reporting honest.
Pre-use checks as a compliance record
An obligation in most jurisdictions, usually recorded on paper and rarely audited.
Interlock the check with the machine start, so it cannot be skipped.
Record completion with time and operator.
Record defects reported, which is the part that matters and which paper systems lose.
Route defects to maintenance automatically.
Audit the completion rate and the defect rate together. A high completion rate with almost no defects reported indicates the check is being clicked through rather than performed.