Monitoring Can Change the Job Before It Removes It
Sensors and dashboards often tighten the task before automation replaces it. Audit what is measured, who can contest it, and what new work the measurement creates.
Monitored-work maps the measurement system and managerial rule; occupational-tasks identifies the duty, exception, and added work that the rule changes.
Measurement can reduce danger or drudgery, and an audit cannot establish intent; it can establish the terms imposed on the work.
The handheld device makes a sound you have not heard before.
A timer appears beside the next stop.
Nobody says automation.
The screen says route, task, status, elapsed time.
Maybe the change is a camera aimed at a work area, telematics in a vehicle, a scanner in a warehouse, activity software on a remote laptop, or a dashboard in a control room.
The device arrives before the explanation.
Do not begin by asking whether it is legal, fair, or permanent.
Write down what it measures.
Then write down what changes when the number moves.
The title on your contract can stay exactly where it was while your job becomes narrower, faster, more standardized, more interruptible, or easier to compare against somebody else’s.
That is job change.
It can arrive long before a layoff date.
The Timer
A sensor is not yet a rule.
A camera can help somebody see a hazard.
Telematics can show where a vehicle is when dispatch needs to find it.
A remote-work log can show whether a system is connected.
Those uses can reduce danger, shorten diagnosis, prevent breakdowns, make routing more predictable, and keep a specialist away from hazardous travel.
Do not deny the useful case.
Then find the rule that turns the signal into a management action.
Does the number change your sequence of work?
Does it set your target?
Does it determine a route, a shift, a pay adjustment, a warning, a score, a staffing level, or who gets called in next week?
The measurement system changes your job when the answer is yes.
The important object is not the device.
It is the chain.
Signal to dashboard.
Dashboard to threshold.
Threshold to action.
Action to your working day.
The business case is rarely a sensor purchase alone.
It includes devices, communications, integration, maintenance, supervision, insurance, training, data governance, and a changed work intensity.
Your employer may call this visibility.
Visibility is not neutral after it is connected to a rule.
Take out a sheet of paper.
Start the register before the system’s labels become the only description of your work.
The Register
Draw three columns.
Task measured.
Write the action in ordinary language.
“Drive from stop to stop.”
“Pick the item.”
“Close the ticket.”
“Complete the lift.”
“Answer the customer.”
Do not write a title.
Customer service is not a measured task.
“Explain the changed delivery time to the customer after the route fails” is closer.
For work that crosses occupations, use the action rather than the title.
A similar action does not establish equal responsibility, setting, training, or exposure to a local score.
Action triggered.
Write exactly what follows the number.
“Prompt appears.”
“Supervisor receives an alert.”
“Route is reassigned.”
“Score falls.”
“Break is questioned.”
“Vacancy is not backfilled.”
Do not supply a motive the rule does not state.
The register cannot establish intent.
It can establish the terms imposed on your task.
Unmeasured work required.
Write the action the system needs but does not properly see.
Walk to the equipment.
Wait for a customer.
Read the unclear instruction.
Check the source record.
Call for help.
Explain the exception.
Rebuild the audit trail.
Recover the customer after the output is wrong.
This is where your job changes first.
A system can remove time from one visible task and create checking, data correction, annotation, maintenance, review, and exception work around it.
The dashboard records a completed transaction.
You still make the transaction safe enough to complete.
What Is Counted
Do not accept productivity improved as a complete sentence.
Ask what is counted.
Completed tasks per paid hour?
Stops per route?
Minutes per lift?
Calls closed per shift?
Exceptions per transaction?
The denominator changes the claim.
Minutes per lift cannot establish that two lifts are comparable.
One may include a defective camera, a new exclusion zone, a load that requires a brake test, a weather delay, or a ground worker’s warning.
The telemetry can look complete while omitting radio instructions, a camera blind spot, or a condition that changes what safe work requires.
Mark each reading as measured, inferred, delayed, or unavailable.
That distinction matters when a signal is stale.
Under OSHA’s construction-crane rule, if electronic rated-capacity information becomes inaccessible, the operator immediately ceases operation or follows a safe shutdown procedure until the information returns.
That is a stopped machine.
It is not an admission of low productivity.
The same rule requires the operator to verify load weight before proceeding when an indicator suggests more than 75% of maximum rated capacity at the longest radius.
At 90% or more of maximum line pull, the operator lifts the load a few inches and tests the brakes.
At least 2 full wraps of rope must remain on the drum.
These are not numbers for a generic target sheet.
They are examples of why a count without its controlling condition becomes a false judgment about a person.
Safety devices and operational aids do not substitute for professional judgment.
Do not let a performance metric claim that they do.
The Single Number
The pressure to simplify will arrive as a score.
Resist the wrong simplification.
O*NET marks a task core when it is relevant to at least 67% of respondents and rated at least 3.0 for importance.
That does not mean it happens every day.
It does not mean it takes most of a shift.
It does not mean it can be automated.
It does not mean a worker in every local version of the occupation performs it the same way.
Relevance, importance, and frequency answer different questions.
A worker survey asks something different again: whether a person used a generic task in a particular job, on a scale from 0, never, to 4, every day.
Neither source gives a manager permission to turn one task number into a complete account of your working day.
An occupational statement says what a duty can be.
A worker report says whether a respondent used a task in a particular job.
Your register says what the local measurement system is doing now.
Keep the units separate.
For a task record somebody can check later, keep the occupation code, the task wording, the source, the collection date, and the rating method.
A one-word label is convenient for a dashboard.
It is weak evidence for a discipline decision.
A score without its collection history cannot settle a close comparison.
Do not confuse a technical-potential score with a work record.
The ILO’s generative-AI assessment rates tasks from 0 to 1, not possible to entirely possible, before aggregating to occupations.
That model output is not a log of adoption, pace, pay, discipline, or staffing at one employer.
Pace Before Headcount
The economic attraction of monitoring is not merely information.
It lets an employer standardize routing, rank work, allocate tasks, set thresholds, and compare sites or people at low marginal cost.
That can expose a bottleneck nobody could previously see.
It can also make walking distance, waiting, equipment failure, care work, safety checks, and hard cases disappear from the price of the task.
A performance measure changes the product being produced, because workers adapt to what it rewards.
Study the adaptation.
Does a dispatcher stop assigning difficult calls because they lower the visible rate?
Does a warehouse worker delay asking for help because the timer keeps running?
Does a remote worker keep a system active instead of doing the off-screen thinking that prevents an error?
Does a vehicle operator avoid a safety stop because the log will call it idle time?
The question is not whether the score is objective.
The question is what behaviour it makes costly.
Do not accept a productivity claim without its period, numerator, denominator, and baseline.
Early apparent savings can be consumed by debugging and training.
Later savings can come from staffing changes, greater output, stricter pace, or work shifted to customers.
Those are different ledgers.
Make your employer identify the ledger: asset downtime, labour hours per unit, service quality, incident cost, payroll, or another defined measure.
Then identify whose work left the first ledger.
Keep the date attached to any claimed saving.
A measurement claim without a period cannot distinguish initial debugging from later staffing change.
Connected systems can move a field operator to remote operation, a clerk to reconciliation, a supervisor to alert response, or your old data entry to the customer.
Reduced labour time is not a complete account of where the task went.
Four Points
Now score the system.
For every new metric, test whether it changes pace, pay, discipline, or staffing, including assignment, shifts, backfilling, or access to work.
Give the metric 1 point if any of those conditions applies.
The count is a trigger, not a validated measure, and it cannot prove what your employer intends.
Total the points across the metrics now in use.
At 4 points, request the record in writing.
Ask for the rule, the denominator, the baseline, the action triggered, the error process, the correction route, the retention period, and who can make an employment decision from the score.
Ask whether the rule distinguishes a defensive pause from inattention.
Ask whether an error can be corrected before pay, discipline, scheduling, or staffing changes.
Ask whether you can see the underlying event, source, timestamp, and status label.
Ask who reviews an exception.
Ask when a personal identifier is no longer needed.
These are operating questions.
They are not requests for reassurance.
Who Owns the Score
The score can outlast the shift.
Ratings, reviews, availability records, and location history can affect your access to work and your future bargaining power, even where no formal wage cut appears.
That is why ownership and portability matter.
The EU platform-work directive gives workers a right to portability of personal data generated through the performance of work, including ratings and reviews, and requires platforms to supply the tools free of charge.
Its rule applies in that context.
The economic mechanism travels farther.
When a reputation cannot move, switching costs rise.
When the record travels with you, the employer or platform is no longer the sole owner of the evidence about your work.
The usable life of a monitoring record is a governance decision, not a natural property of a dashboard.
Governance is part of the price of the system, not a decorative compliance layer.
Data inventories, access controls, retention review, testing, notices, human review, appeal handling, and periodic evaluation cost something.
So do investigation, turnover, litigation, rework, and lost trust when those things are omitted.
New York City requires a covered automated employment decision tool to have a bias audit within 1 year of use, with the results made public.
That is not a universal fairness guarantee.
It is a reminder that a vendor quote can hide the cost of evaluating a system that allocates valuable work.
The person carrying the score carries that hidden cost first.
The Work It Creates
Do not let the dashboard erase the workers who make it usable.
A closed-loop system does 4 linked things: it monitors, analyses, manages, and simulates a real asset.
Each of those needs a model, a data pipeline, computing, a user interface, and operational authority.
Visualisation is not closed-loop control.
Closed-loop control adds validation, cyber-security, and change-management work.
A remote or semi-remote system also needs a safe failure state.
OSHA describes manual robot operation as inhibited unless a 3-position enabling device is held in its centre position.
The device is not the whole safety cost.
Testing, procedures, trained operators, and records are part of it.
Write these tasks in your third column.
Not because they guarantee you a job.
Because they are the work the first savings claim is likely to omit.
Record the review task.
Record the exception task.
Record the new hand-off.
Do not pretend a task invented last quarter has the same standing as a stable duty.
But do not let the title hide it either.
Minimum Viable Action
Before you sleep tonight, list every metric your work device recorded today.
Beside each one, write the task it measured.
Then write the action it triggered.
Then write the work it did not record.
Count the points.
At 4 points, request the rule, error process, and baseline in writing.
Do not wait for the word automation.
What you will be able to say afterwards
- A monitoring metric changes the job when it changes pace, pay, discipline, or staffing.
- A three-column register connects a measured task to the action triggered and the unmeasured work required.
- Safety telemetry and productivity control require different interpretations of the same event log.
- A metric needs a disclosed rule, denominator, baseline, error process, and contest route before it can support an employment decision.
- O*NET core status, relevance, importance, and frequency describe different task properties and cannot be collapsed into one score.
- Monitoring often creates review, annotation, exception, and recovery work before it removes headcount.
Questions this raises
Does a new timer mean that the job is being automated?
No. A new timer establishes that an employer can record a duration and compare it with a rule. It does not establish that a task can be automated or that a job will disappear. The job changes when the recorded duration triggers a pace target, pay consequence, discipline, staffing decision, route allocation, or other managerial action. The relevant evidence is the rule attached to the timer.
What should go in a monitoring register?
A monitoring register has three columns for each new metric: the task measured, the action triggered, and the unmeasured work required. The first column names a concrete duty rather than a job title. The second records the manager, system, or policy response. The third records safety checks, walking, waiting, repair, explanation, escalation, customer recovery, or other work that the metric leaves out.
When should a worker ask for the monitoring rule in writing?
A worker scores one point for every metric that changes pace, pay, discipline, or staffing. At four points, the worker requests the rule, the error process, and the baseline in writing. The score is a structured prompt rather than a validated instrument. Its purpose is to distinguish a passive record from a measurement system that is altering employment terms.
Why can a safety pause look like poor performance on a dashboard?
A dashboard can record a stop without recording why it occurred. OSHA requires a crane operator to stop operating or follow a safe shutdown procedure when electronic rated-capacity information becomes inaccessible. That is a safety response, not a performance failure. A usable event log distinguishes defensive pauses, stale data, sensor faults, and site conditions from delay attributed to a worker.
What does the 75% crane threshold require?
Under OSHA's construction-crane rule, when an indicator suggests that a load exceeds 75% of maximum rated capacity at the longest radius, the operator verifies the load weight before proceeding. The sensor reading is an input to a safety decision. It does not authorize a dashboard to replace the lift plan, the accepted source, or the reliable calculation used to establish the load.
Can a productivity measure treat brake testing as lost time?
It should not. OSHA requires a brake test when a load reaches 90% or more of maximum line pull: the load is lifted a few inches and the brakes are applied. The time is real work required to operate safely. A productivity measure that records only minutes per lift can misclassify that protective action as individual inefficiency and reward its avoidance.
What is the difference between a core task and a frequent task?
O*NET defines a core task as relevant to at least 67% of respondents and having mean importance of at least 3.0 on its rating scale. That classification does not mean the task occupies most of a shift or occurs most often. Frequency, relevance, and importance are separate attributes. An infrequent safety or legal decision can be central even when a routine administrative action happens repeatedly.
How should a worker record a task so the record can be checked later?
A checkable task record keeps the occupation code, task ID, wording, task type, release version, update date, source, and any rating method. O*NET's task-statement table has eight named fields, and task wording can be as long as 1,000 characters. A short title or a single verb is not enough because it can detach an action from its object, purpose, and occupational context.
Do sensors always make work worse?
No. Connected monitoring can reduce hazardous travel, improve diagnosis, prevent breakdowns, make routing more predictable, and allow specialist support across locations. It can also standardize work intensity, rank people, and reuse a safety record for discipline. The effect depends on the rule, the error trade-off, the task, and the action triggered by the measure; a device purchase alone does not settle it.
What happens to the work that a dashboard does not record?
Unrecorded work does not disappear because it lacks a field. It can become review, data correction, annotation, exception handling, maintenance, customer recovery, or a colleague's unpaid adjustment. A monitoring system can show movement, location, and elapsed time while omitting radio instructions, a ground worker's warning, a camera blind spot, or the reason a task could not safely proceed.
What does a 0 to 4 task-use answer measure?
In PIAAC work-task questions, 0 means never and 4 means every day. The responses record reported task use in a particular job, including properties such as incidence, variety, frequency, complexity, and criticality. They do not establish that every worker with the same occupation code performs the task at the same rate or that reported frequency is a measure of ability.
Can a worker challenge an automated employment score?
The available process depends on the jurisdiction and employer, but the score needs an identifiable rule, data source, error route, and human decision point before it can be meaningfully contested. New York City requires a covered automated employment decision tool to have a bias audit within one year of use and public information about that audit. That jurisdiction-specific rule is not a universal fairness guarantee.
Why does ownership of ratings matter after a worker leaves?
Ratings, reviews, availability records, and location histories can affect access to shifts and future bargaining power even without a formal wage cut. Directive (EU) 2024/2831 gives people in its platform-work context a right to portability of personal data generated through work performance, including ratings and reviews, with supporting tools provided free of charge. Portable evidence can reduce switching costs; a proprietary score can increase them.