Checking a payslip against your own record

Most people glance at the net figure, decide it looks about right, and file it. A proper check takes ten minutes, and the first time it finds something it pays for every month you have ever spent doing it.

Before you start: know your cut-off

Find out the period your payslip actually covers. It is frequently not the calendar month. A cut-off on the 20th or 25th is common, as are four-weekly and week-based cycles.

This one fact prevents most false alarms. Under a cut-off on the 25th, everything you work between the 26th and the end of the month appears on the next payslip. Comparing a calendar-month total against that will show a difference every single month, and it will not be an error. Establish the cut-off once, write it down, and adjust your comparison window to match it.

The ten-minute routine

Step 1 — hours, not money

Compare the hours figures first, split into normal and extra. Money is a function of hours and rates; if the hours are wrong, the money conversation is premature and will confuse the issue.

Three outcomes. Both match: go to step 2. Total matches, split does not: classification problem, go to step 3. Total does not match: go to step 4.

Step 2 — rates

For each pay line, divide the amount by the hours. That gives the rate actually applied. Check it against what you expect: base rate for normal hours, base × multiplier for overtime, base plus premium for unsocial hours.

A rate that is wrong is almost always wrong every month, which makes it worth finding precisely once. Note also whether allowances are inside or outside the calculation base for premium rates — contracts differ, and it changes the expected figure.

Step 3 — classification

If total hours agree but the normal/extra split does not, hours you recorded as overtime were paid as normal time or the reverse. The usual causes are a daily threshold that differs from what you assumed, a weekly rather than daily threshold, or overtime that required authorisation and did not get it.

This is worth resolving properly rather than month by month, because the same misunderstanding will recur. Ask for the rule in writing.

Step 4 — find the missing days

When totals differ, divide the difference by your usual shift length. If the result is close to a whole number, you are looking for that many whole days — a shift that was never submitted, a day recorded against the wrong job, a duplicate entry on your side.

If the difference is small and does not divide cleanly, it is usually breaks: a scheduled deduction applied where you recorded an actual one, or unrecorded time at the edges of shifts. Those are covered in breaks and unpaid time.

Step 5 — the non-hours lines

Then check what is not driven by hours at all: allowances, bonuses, meal vouchers, expense reimbursements, deductions. These are easy to skip and easy to get wrong, because they are often entered manually by a person rather than computed by a system.

Also check absence lines. A sick day paid at the wrong rate, or annual leave deducted from the wrong balance, shows up here rather than in the hours.

Step 6 — the hour bank

If you have a banked-hours balance, check the movement on it: hours credited this period, hours drawn, closing balance. Bank movements are frequently outside the payroll system entirely, which is exactly why they get lost. See the hour bank guide.

Compare gross to gross. Your record produces a gross figure. Net pay differs by tax, social contributions and any other deductions, none of which your tracker knows about. Every "my record says 2,400 and the payslip says 1,780" query is this.

When you find a real difference

How you raise it matters as much as whether you are right.

Keep the checks

Note the outcome of each month's check somewhere — even just "checked, matched". Two things come out of this. A recurring error becomes visible as a pattern rather than as an isolated annoyance, which changes how seriously it gets treated. And if a dispute ever escalates, a consistent history of monthly checks makes your record considerably more credible than one assembled after the fact.

Most months, everything matches and the ten minutes feels wasted. That is the normal case, and it is not wasted — it is what makes the exception detectable.

Last updated