Payroll Register

How does a register differ from a payslip?

By audience and by scope. A payslip is one person's view of one period, written to be understood by that person without help. The register is the employer's view of every person in that period, written to be reconciled by somebody who does this professionally. Each payslip can be derived from the register; the register cannot be derived from a stack of payslips, because it also carries totals, employer-side amounts and the classifications that the accounts and the statutory filings are built on. The two documents serve different readers and should not be expected to look alike or to contain the same fields.

How does it differ from the accounting entry?

The register records what was calculated for each individual, name by name. The accounting entry records the cost by category and cost centre, aggregated, in the ledger's structure rather than the payroll system's. One is a list of people; the other is a set of balances, and neither can be read as the other without a translation. They should reconcile, and the mapping between them deserves to be documented and owned, because when the two disagree the argument is nearly always about how a component was mapped rather than about what anybody was actually paid that month.

Why must a register be frozen?

Because it is the evidence that a particular run happened in a particular way. If the underlying records remain editable after the run, then re-running the report tomorrow produces a different register for the same period, and neither version can be relied on by anybody who was not present at the time. Freezing the calculated output at the point of approval - and keeping it unchanged alongside every later correction rather than instead of it - is what allows the run to be reconstructed months afterwards by somebody who has to explain it and was not there.

What is the register actually used for?

Four purposes, and they pull the document in slightly different directions, which is why registers designed for only one of them disappoint the others. Reconciliation compares the total calculated against the total disbursed and against what was posted. Statutory reporting draws the amounts and classifications that go to the authorities. Accounting takes the cost split by category and cost centre. Analysis asks what changed against the previous period and why, which is where the register stops being a compliance artefact and becomes management information that somebody outside the payroll team wants to read and act on.

Serving all four means the register has to carry considerably more than net figures. Employer-side contributions, the classification each component falls into, cost centre and location, and the identifiers used in external filings all belong in it, even though none of them appear on any individual's statement and no employee would recognise them. Where the register is treated as a bulk list of payslips, the reconciliations and the filings end up assembled from separate extracts pulled at different moments, and those extracts diverge from one another over time without anybody noticing until two of them are compared.

What should be read first on a register?

Not the largest numbers, which is where most people start. Begin with the population: does the count of people paid match the count expected, and does the difference reconcile to the joiners and leavers of the period? A person paid who has left, or a joiner missing entirely, is both the most consequential error available and among the easiest to find, and neither surfaces in a review that begins with amounts. Only then look at anybody whose net moved sharply against the previous run, in either direction, and establish what changed for each of them before releasing anything.

After that, the tails, where the unusual entries live. Zero or negative net amounts usually signal a deduction exceeding earnings, which is worth understanding before the payment goes rather than afterwards from the employee. Duplicate bank accounts across different employees deserve a look, since there are legitimate explanations and one unpleasant one. Amounts far outside the range for a grade or location may well be correct and are worth confirming anyway. This sequence takes minutes on a register with the right sorts available, and it catches the errors employees would otherwise find first and report loudly.

Where do register-to-ledger differences come from?

Timing accounts for most of them. A payment made outside the run, a remittance falling in a different accounting period, an accrual raised for something the run has not yet paid - each produces a real and explainable difference between what payroll calculated and what the accounts show, and each is entirely correct. Rounding accounts for a smaller and more persistent share that never quite disappears. Neither is a problem for as long as every difference is identified and described at the time it arises, by somebody who knows what it was rather than somebody guessing later from the size of it.

The differences that matter are the ones nobody can name. An unexplained variance in a payroll reconciliation is usually a mapping error, a component posted to the wrong category, or a run whose output changed after posting, and all three become harder to trace the longer they sit unexamined. Reconciling every period rather than at quarter end keeps the search space small enough to work through. Where [HR analytics](/hr-analytics-software) reporting also draws on payroll data, a third figure enters the picture and should be reconciled to the same frozen register rather than to a fresh extract taken on a different day.

Who should be able to reach the register?

The register is the single most sensitive routine document most organisations produce: everybody's pay, on one screen, sortable by amount. Access should be granted explicitly rather than inherited from a system role somebody was given for an unrelated reason, and the list of who holds it should be short enough to recite from memory. Requests from managers wanting to see their team's costs are entirely reasonable and are better met with a filtered view than with the whole file, because the whole file is the thing that leaks and cannot be recalled once it has been forwarded.

Export is where control is usually lost, quietly and with good intentions. A register downloaded into a spreadsheet for a legitimate reconciliation stops being governed by the [payroll management](/payroll-management) system's permissions and starts being governed by whoever holds the file and wherever they saved it. Where exports are necessary, agree where they are stored, who may open them and when they are deleted, and prefer reporting inside the system for anything recurring. A one-off export that quietly becomes a monthly habit is how a confidential document ends up in a shared folder nobody remembers creating.