Corrections and What They Must Show

Working time records are amended routinely: a missed clock-out, a shift that ran over, a correction to a job record, a manual entry for somebody who was at another site. Most amendments are legitimate and necessary.

The evidence described in “Corrections and What They Must Show” should preserve the original entry, the correction and the approval instead of presenting only a final number. A team assessing time tracking software with accountable controls for time tracking software should run that full sequence during a pilot and confirm that exports remain understandable without access to the live account.

What matters is what the record retains about them, because an hours figure that can be changed without trace is not a record of hours. It is a current value.

For an independent reference relevant to “Corrections and What They Must Show”, consult the Harvard Business Review management resources. Use it to test working-time definitions, recordkeeping, access, retention and exception handling against the organisation’s real process rather than treating one software report as conclusive.

What an amendment should capture

The original value. The new value. Who changed it. When. And why, in a free-text field somebody actually filled in.

Five fields, and the fifth is the one that decides everything. "Correction" as a reason is the same as no reason. "Clocked out at the terminal by the loading bay, system shows no punch, supervisor confirms 18:40 finish" is a reason.

The sequence, not the last state

Amendments get amended. A system that keeps only the current value and one previous has lost the first correction and its reason.

What is needed is every state the record has held, in order, each with its author and reason. Most systems can do this and it is frequently switched off, usually for storage or to simplify a report. It is the single setting most likely to determine whether the data supports a contested conclusion.

Who may amend

A short list, named, with the ability recorded in the system rather than in a policy. Supervisors for their own teams, payroll for corrections after the period closes, and nobody else.

The list matters less than the fact that it exists and that amendments carry the name of the person who made them. An organisation where any administrator can change any hour has records whose weight is correspondingly low.

Amendments after the period closes

These need separate treatment and a higher bar, because they change a figure that has already been used — for a payroll run, for a compliance report, or in a document given to somebody.

Require an additional approval and a stronger reason. Then check periodically whether the volume is rising, because a growing count of post-close amendments usually means something upstream is broken rather than that more mistakes are being made.

The direction of amendments

Worth monitoring as an aggregate: do corrections tend to increase or decrease recorded hours, and by how much.

A systematic bias in one direction is a finding regardless of which direction it is. Downward suggests a pressure on the figures; upward suggests the capture is missing hours routinely. Either way it is better for the organisation to notice than for somebody else to.

Volume as a diagnostic

Corrections per hundred shifts, by department, by month.

A high rate in one department points at something specific: a terminal in the wrong place, a shift pattern the system handles badly, a supervisor using corrections where another mechanism exists. Each has a fix, and the rate is how you find out which one you have.

Amendments to the compliance figures themselves

A different and more serious category: changing the stored average, the opt-out date, or a rest record.

Those should not be amendable at all except through a documented process with a named approver. Where a system permits silent editing of compliance data, that is worth raising as a defect, because the existence of the capability weakens every figure the system produces.

The five-minute check

Pick an amended record from a year ago. Ask for the original value, every subsequent state, the author of each change, and the reason given each time.

If all of that comes back in five minutes, the process works. If the system shows only the current figure, the organisation's working time records cannot support any contested conclusion at all — which is worth knowing before the first contest rather than during it.

Correcting a correction

Amendments get amended, and a system holding only the current value and one previous has already lost the first change and its reason.

What is needed is the full sequence. Most products can keep it and many installations have it switched off to save space or simplify a report. It is the single setting most likely to decide whether the records support a contested conclusion, and turning it back on costs nothing but the storage.