Rounding

Rounding is a legacy of paper timesheets that survives in software defaults. It is the single easiest place to find unpaid time.

The three settings

Increment: the unit — six minutes, a quarter hour, the minute.

Direction: up, down, or nearest.

Threshold: where nearest tips.

Products conflate these, and a configuration described as "round to fifteen minutes" tells you only the first.

Where it goes wrong

Rounding down on arrival and up on departure is the classic finding, and it is a default in some systems rather than a decision by anyone.

Rounding applied to breaks as well as shift boundaries, compounding the effect.

Rounding at capture rather than at calculation, which destroys the raw value and makes the test below impossible.

The test

Ten minutes, and it settles the question.

Sum the raw recorded minutes for a period.

Sum the rounded minutes for the same period.

Compare.

Neutral rounding produces a difference near zero.

A consistent gap in one direction is the finding, and the direction tells you whose minutes they are.

Run it per site, because a configuration applied to one terminal and not another shows up here.

What the rules generally require

General orientation; check your jurisdiction.

Where rounding is permitted at all, neutrality over time is the usual requirement.

Several jurisdictions have restricted or effectively ended it, and the direction of travel is against.

The safest position is not to round. Systems record to the minute, and the increment is a billing or payroll convention that can be applied at calculation.

Fixing it

Store raw, round at calculation, which keeps the test possible and lets a policy change apply without a migration.

Set direction to nearest, or remove rounding entirely.

Check breaks separately, because they are frequently configured differently and nobody looks.

Then re-run the test and keep the result, because a documented neutrality check is worth having if anyone asks.

Why this belongs in a time-theft collection

Because it is the mirror image.

An organisation investigating minutes taken by workers, while rounding takes minutes from them by configuration, has an inconsistency that will be raised — and the arithmetic proving it takes ten minutes for anyone who asks.

Keep the result

A documented neutrality check is worth having.

Run the ten-minute test, record the date and the figure.

Repeat quarterly and after any system update, because configurations drift and defaults return.

If anyone asks whether rounding favours the employer, a dated series of near-zero results answers it — and an absence of any check answers it differently.

Connect policy and data

The choices in this note can be compared with open this project connection. Keep the written purpose in control and enable only the information needed at this boundary.

Independent reference

For a thematic point of reference, see the U.S. Department of Labor. Use this established source as an outside check before turning the principle into a system rule.