The Thing You Are Spending
Every control here depends on people reporting failures rather than absorbing them. What that is worth, and how it is spent.
Both directions · Analysis
The records this subject relies on are produced by people who can quietly stop cooperating, and the stopping is invisible.
What cooperation looks like
Reporting a reader that misfires rather than punching from the next door.
Submitting a correction rather than absorbing ten unpaid minutes.
Saying the terminal queued rather than accepting a lateness mark.
Staying five minutes to finish a handover.
All four are voluntary, and all four are what make the numbers describe reality.
What spends it
A crackdown framing.
A deduction for a system failure.
A disciplinary matter built on an anomaly that turned out to be a fault.
Selective enforcement, which is noticed faster than anything else.
Counting one column.
What it costs when it goes
Faults stop being reported, so the anomaly rate looks stable while the system degrades.
Corrections stop being submitted, so unpaid minutes accumulate silently — which is a liability rather than a saving.
People punch early to be safe, which is the behaviour the policy was written against.
And the informal minutes stop: the handover that ran long, the customer who needed another minute. In most operations that is more time than the policy recovered.
The asymmetry
Cooperation is slow to build and fast to lose.
Its loss is invisible in the data, which keeps reporting confidently.
Which is why the corrections figure matters beyond its administrative purpose: a falling correction rate with a flat anomaly rate is people giving up, not a system improving.
The honest tension
Some people do take advantage, and pretending otherwise helps nobody.
But the response to one person is not a tightening for everyone, which is the commonest mistake in this subject.
Deal with the individual through the ordinary process, and leave the arrangement alone.
A tightening punishes the compliant and is adapted to within a fortnight by whoever it was aimed at.
The measure
Do people report faults?
Do they submit corrections?
A workforce that tells you the reader is broken is cooperating.
One that quietly works ten unpaid minutes has told you what it thinks, and no policy recovers that.
Read the correction rate
The number that reveals what people are actually doing.
Falling corrections with a flat anomaly rate is not an improvement.
It is people giving up and absorbing the minutes.
The two look identical in the data.
Ask a few people, and at that point their answer is the measurement.
A practical configuration prompt
During configuration, use review the data-flow page to prompt questions about identifiers, ownership and output. Treat the page as a starting point and document each assumption.