When Somebody Says the Record Is Wrong

Somebody says they worked hours they have not been paid for. It arrives as an email to payroll or as a conversation with a supervisor, and what happens in the first forty-eight hours decides how long it takes.

The correction process in “When Somebody Says the Record Is Wrong” needs a visible history rather than an overwritten final number. If this practical implementation page supports 7 minute rule payroll, administrators should test amendment, approval and export workflows in advance so a later dispute can distinguish the original entry, the question asked, the correction and the person who accepted it.

The common mistake is to answer from the punch report. The punch report is the organisation's position, not its evidence, and in most of these disputes the punch report is exactly what is being disputed.

For a separate benchmark relevant to “When Somebody Says the Record Is Wrong”, consult the GitLab remote-work handbook. Use it to test record quality, approvals, retention, employee rights and exception handling against the real workflow rather than treating a software report as self-explanatory evidence.

The first move

Gather before responding. Punch data for the period, the correction history for those punches, access-control events, any paper records, the roster as published and as amended, and the supervisor's own notes if they exist.

Most of that is available within a day and some of it will contradict the punch report, which is better found now than after a position has been stated.

An organisation that denies a claim and is then contradicted by its own access data is in a far worse place than one that took two days.

What usually turns out to be true

The person worked the hours and the punch does not show them, because of the edge described throughout this collection: they were on site, doing something required, before or after the punch.

Or a correction was made that reduced their recorded time and they were not told.

Or the terminal failed on one of the days and the paper record is incomplete.

Deliberate fabrication by the employee is rare in this category — rare enough that treating the claim as presumptively false is both wrong and expensive.

Who should handle it

Not the supervisor whose records are in question, and not the person who made the correction.

Someone with access to all the systems and no stake in the answer. In most organisations that is payroll with HR support, and the useful thing about that pairing is that neither owns the operational decision being examined.

Proportionality

Most of these claims are small: a few hours, a few hundred in pay. The cost of a contested process exceeds the amount several times over.

Where the evidence is ambiguous and the sum is modest, settle it, in writing, with a note of the reasoning. That is not an admission about anything else and it is considerably cheaper than being right slowly.

Reserve the full process for claims that are large, repeated, or where the facts matter beyond the individual — which is the case where several people are making the same claim.

When several people raise the same thing

The threshold at which this stops being an individual matter. Three claims of the same shape about the same arrangement is a finding about the arrangement.

It also changes the exposure: limitation periods in most jurisdictions reach back further than people assume, and a practice running for years across a site is a multiple of what the current claims suggest. The moment a second claim of the same shape arrives, the question is whether the practice is still running.

Representation

Where there is a union or a representative, engaging early is almost always faster. They have usually already gathered the pattern and they know how many people are affected, which the employer does not.

A dispute about a boundary is a collective question wearing an individual costume, and treating it collectively resolves it in one conversation rather than in twenty.

The document that prevents most of this

The boundary document described earlier in this collection, written before any dispute, stating which activities are inside the paid day at this site.

Where it exists, these claims are resolved by reference to it in an afternoon. Where it does not, every claim is argued from first principles, and the organisation argues each one without knowing what it decided.

What it costs to be wrong about this

The arithmetic of a contested hours claim is uncomfortable. The disputed sum for one person over a year is usually modest; the cost of defending the position across a population, with the same configuration behind every case, is not.

That asymmetry is the reason to resolve the single case on its merits rather than on principle. A claim settled because the record was weak is cheap; the same claim defended to a finding that the configuration was wrong applies to everybody on that configuration, and the organisation will have chosen the expensive version of the same conclusion.