Hours Recorded Against Hours Paid

The time system holds hours recorded. Payroll holds hours paid. They are produced from the same punches by the same chain and they disagree, for reasons that are individually explicable and collectively unexamined.

The record described in “Hours Recorded Against Hours Paid” should be created close enough to the work that people are not reconstructing a polished week from memory. When teams assess the product website for key person dependency, they should keep entry, project selection and correction simple, while explaining which optional activity data is collected and how employees can review it.

Nobody reconciles them because each system is internally consistent, each department trusts its own output, and the comparison belongs to neither.

For a separate benchmark relevant to “Hours Recorded Against Hours Paid”, consult the ACCA technical resources. 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.

Why they disagree

Manual adjustments in payroll that did not go back to the time system: a correction, an advance, a claim processed by exception.

Records added in the time system after the extract was taken, which is the most common and is a timing problem rather than an error.

Elements that exist in one and not the other: an allowance paid in payroll with no corresponding time record, or a recorded category that maps to nothing.

Rounding applied in both, which happens more often than anyone expects when two systems each tidy the figures.

And genuine errors in the chain, which are the minority and the reason to look.

Running it

Export hours by person from the time system for the period. Export paid hours by person from payroll. Join on the payroll number.

Three buckets: identical, differ by a small tolerance, differ materially. Set the tolerance deliberately rather than by what makes the list short.

Half a day for a site of a few hundred, less once the recurring categories are known.

What to do with the material differences

Each needs a cause, and the causes repeat. After the first run there will be four or five categories and most of the list falls into them.

Underpayments get corrected promptly and the person is told. This is both the right thing and, in most jurisdictions, an obligation that does not depend on them having noticed.

Overpayments are raised promptly, because recovery is constrained by time and amount nearly everywhere and the conversation at one month is far easier than at nine.

And the category that caused the most differences gets fixed upstream, which is the point of the exercise.

The pattern rather than the instances

A reconciliation that finds thirty differences and corrects thirty records has done a third of the job.

The useful output is which category dominates. Mostly timing means the extract schedule is wrong. Mostly manual adjustments means payroll is working around something. Mostly unmapped elements means the configuration has drifted from the agreement.

Each is one fix that removes a recurring class rather than an instance of it.

Doing it monthly

The first run is the hardest and finds the most. By the third the recurring categories have been fixed and the job takes an hour.

That is the level at which it becomes sustainable, and sustainability is the whole point: an annual reconciliation finds a year of accumulated differences at once and is usually abandoned.

Who should own it

Somebody who can see both, which in practice means payroll with an agreed route into the time system, or a workforce analyst who sits between.

It should be a named standing task with a date, not an improvement project. Tasks on a calendar survive a change of personnel; projects do not.

The side benefit

The reconciliation produces, free, the list of manual entries and corrections for the period — who made them, how many, and against which people.

That list is the input to the correction monitoring described later in this collection, and an organisation doing the monthly reconciliation already has it without any additional work.

The first month is not the test

A first reconciliation produces a large number of differences, most of them explainable, and it is tempting to conclude from the volume that the exercise is unworkable.

The useful signal is the second and third months, once the structural causes have been described and excluded. A set that shrinks is a process coming under control; a set that stays the same size with different names in it is a process that is not. Only the second case needs work, and nobody can tell which they have from a single month.

Keep the excluded structural causes written down as well, because a cause that was explained once tends to be rediscovered every few months by somebody new to the report.