What the Records Will Support

The hours data ends up supporting conclusions it was never designed for: a disciplinary case about attendance, an insurance claim after an accident, a dispute about unpaid overtime, a client's audit of a supplier's practices.

The evidence described in “What the Records Will Support” should preserve the original entry, the correction and the approval instead of presenting only a final number. A team assessing open the official page for employee monitoring for performance reviews should run that full sequence during a pilot and confirm that exports remain understandable without access to the live account.

Each of those asks the record to bear weight. How much it can bear depends on provenance, and the system presents every figure as an identical row in a table.

For an independent reference relevant to “What the Records Will Support”, consult the Cloudflare zero-trust overview. 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.

The hierarchy of entries

A timestamp captured at the time by a device: strong evidence of when a credential was presented somewhere.

A rostered figure: evidence of what was planned, not of what happened.

A manual entry made by the person: a contemporaneous self-report, which is reasonable evidence and is not a measurement.

A manual entry made by somebody else, after the fact: testimony about another person's hours.

And an amended value with no reason recorded: close to nothing.

Why the distinction disappears in reports

Because the standard export is a column of hours. The method is held somewhere, usually in a field the report does not include.

So a bundle prepared for a hearing contains forty numbers of four different kinds presented identically, and whoever prepared it was not being careless — they exported what the system offered.

The one change that fixes it

Surface the source on every entry, as a visible column on every report that leaves the system: clocked, rostered, self-entered, entered by another, amended.

It costs a change to a report specification. It prevents the specific failure where a figure is relied on more heavily than it can bear, and it is almost never configured.

The gaps, which are also evidence

A week with no records is not evidence that nobody worked. In a dispute about unpaid hours, an absence of records tends to work against the employer, because the obligation to keep them sits with the employer.

That is worth understanding before relying on a gap. An organisation arguing that somebody did not work hours it has no record of is making an argument about its own record-keeping.

Accidents and the fatigue question

After an incident, the hours worked in the preceding period will be asked for, and the question is about rest more often than about totals.

An organisation that can produce the daily rest position for the person for the preceding fortnight is in a very different position from one that can produce weekly totals only. This is the single most consequential use these records get, and it is the one that depends on the data most organisations purge.

Unpaid overtime claims

The claim is usually that hours were worked beyond those recorded, and the evidence on the other side is frequently messages, documents and system access logs showing activity.

Where the organisation's own records show only what passed through a clock, those other traces will be the better evidence, and they will not be the employer's. Capturing the uncounted categories is therefore not only a compliance exercise; it is also what puts the organisation's records on the same footing as everybody else's.

The sentence to put on any export

Something like: this report shows hours recorded by the method stated against each entry; entries marked as amended or manually entered were not captured at the time.

Two lines on the front of a bundle. They prevent a reader assuming uniformity that does not exist, and they are the difference between records that are candid and records that will be characterised as misleading by somebody else.

What to do with all of this

Audit one bundle. Take a report that was prepared for an external purpose in the last year and check how many of its entries were clocked, rostered, manual or amended.

Most organisations have never looked. The answer tends to be that a quarter of the figures in a document presented as a record of hours were somebody's later reconstruction, which is worth knowing before the next one is produced rather than after.

The bundle prepared under pressure

Most of the harm in this area comes from documents assembled quickly for a hearing by somebody who was not thinking about provenance. They export what the system offers and present it as a record of hours.

A standing instruction helps: any working time data leaving the organisation goes out with the method column and the covering sentence. One line in whatever procedure governs disclosure, and it removes the single most common way these records are overstated — by accident, by people acting in good faith.