Monitoring business cases are built on licence and sensor cost and surprised by everything else.
The visible costs
Licensing, by site, by device or by data volume, and the model matters at scale.
Sensors and tags, including the replacement rate, which is higher in an industrial environment than any datasheet assumes.
Infrastructure for positioning: anchors, cabling, power, mounting at height, survey.
Devices, where handhelds or wearables are involved.
The costs that surprise
Integration effort, mostly your own people's time, and the largest first-year cost.
Data quality remediation, which is a process project rather than a technical one and takes months.
Tag and sensor attrition. Devices in a warehouse get dropped, crushed and lost.
Calibration and re-survey as racking and stock change, which is ongoing.
Ownership. Someone maintains this permanently or it decays within a year.
Consultation and change management, where individual measurement is involved.
The cost that does not appear in the budget
Cooperation, which is real and shows up as data quality.
Where monitoring is read as adversarial, scan discipline degrades, exceptions go unreported, and near-misses stop being logged.
The system continues to produce data, which is why this cost is invisible in the monitoring figures and visible in the injury and turnover ones.
Budget the consultation and the purpose limitation work as part of the project, not as an optional extra, because the alternative cost is larger and harder to reverse.
Sizing the value honestly
From your own elapsed time decomposition, not from a vendor percentage.
Take the largest non-value component — usually waiting or travel — and estimate a conservative reduction.
Convert to hours and then to cost, at your rates.
Show the assumption, so it is the assumption that gets challenged rather than the conclusion.
Separate realised from projected. A saving that has not appeared in the ledger is an opportunity, and conflating the two destroys credibility with finance permanently.
The safety case separately
Do not bundle safety into the productivity number.
The safety case is regulatory and ethical, and it stands on its own without a return figure.
Bundling weakens both: the safety argument becomes contingent on the productivity claim, and the productivity claim inherits the safety argument's unfalsifiability.
What to commit to
Not a productivity percentage.
A specific change in a specific measure, with a date and the same method afterwards.
"Median waiting time between picking and packing from thirty-eight minutes to under twenty within two quarters, measured from the same timestamps over comparable periods."
Checkable, which is uncomfortable and is what makes the next phase fundable.
The phasing argument
Value should arrive before the full cost does.
The first ninety days cost almost nothing and produce findings.
Sensor purchases come after the gaps are identified, and can be phased by zone.
A programme that spends the whole budget before producing a finding has no evidence to justify its second year.
Modelling tag and sensor attrition
A recurring cost that no proposal includes and that surprises everyone in year two.
Devices in a warehouse get dropped, crushed, run over and lost.
Ask the vendor for attrition rates from comparable industrial deployments, not from office ones.
Pilot long enough to measure your own rate, which is the only reliable figure.
Include replacement in the recurring cost, not as a contingency.
Include the labour to replace them, which for ceiling-mounted anchors means access equipment.
Phasing so value arrives first
A programme that spends its budget before producing a finding has no case for year two.
The first ninety days cost almost nothing and produce the decomposition, the constraint and one visible fix.
Sensor purchases come after the gaps are named.
Phase by zone, so a smaller purchase proves the value before the larger one.
Report the finding from each phase before committing to the next.
This also protects against the common outcome where the platform is fully deployed and the operation is unchanged.