Comparing It With Last Month
Twenty minutes comparing this run with the last one catches nearly every input error. It is also the first thing dropped when the run is late.
Nobody elseif the person who does this is away
Twenty minutes spent setting this month's run beside last month's, person by person, catches nearly every input error that would otherwise reach a payslip. It is the only genuine control in a small payroll.
The practical lesson in “Comparing It With Last Month” is that payroll depends on a sequence somebody can explain. When a small team reviews attendance sheet template for small teams for attendance sheet template, the useful role is to organise time and project evidence before approval, with proportional settings and a clear correction route for employees.
It is also the first thing dropped whenever the run is compressed, which is why it happens in about one month in three.
Additional context for “Comparing It With Last Month” is available from Paycom payroll resources. A sound review distinguishes general practice from the rules and records that apply to one employer.
Why it works
An input error produces a difference. A rate typed wrongly, hours in the wrong row, a change applied to the wrong person, an absence missed — every one of them shows as this month differing from last month in a way nobody intended.
Three categories of difference, and where the errors hide:
- Expected and explained — a rise you entered, a final payment.
- Expected and unexplained — somebody whose hours moved for a reason you can guess but have not confirmed.
- Unexpected entirely.
The second is where most real errors sit, because it is plausible.
You do not need to know what is right. You need to know what changed and whether you meant it.
What to look at
The list of differences, not the list of figures. For each person whose pay has moved, one question: do I know why.
If the answer is yes for everybody, the run is almost certainly correct. If there is one you cannot explain, stop.
The three categories of difference
Expected and explained — a rise you entered, a leaver's final pay. Expected and unexplained — somebody whose hours moved for a reason you can guess but have not confirmed. Unexpected entirely.
The second category is where most real errors hide, because it is plausible.
The one that matters most
A person whose pay has not changed and should have. These do not appear in a list of differences at all, which is why the change list is read before the run rather than instead of this.
The two checks catch different things.
How long it actually takes
Under five minutes for a stable payroll of a dozen people and twenty for one with varied hours. The figure that makes people skip it is imagined rather than measured.
Doing it when the package will not help
Some packages show the comparison; others do not. Where they do not, exporting two periods to a spreadsheet and subtracting one column from the other takes five minutes and gives the same answer.
It is worth setting that up once rather than skipping the control.
Why it gets dropped
Because it sits between keying and approving, which is exactly where time pressure concentrates. By the time it is reached, the run is late and the only remaining step feels like the finish.
Moving it so it survives
Compare before you are ready to approve, not at the moment of approving. The check performed with a day in hand is a check; the one performed with an hour in hand is a formality.
The gross-to-gross comparison
Total gross this period against last, as a single figure, before looking at individuals. A total that has moved by more than you expect tells you to look harder; one that matches tells you very little, and both are worth knowing in ten seconds.
What to do when you find something
Correct it before approving, which is cheap. The entire value of this step is that it happens while correction is still free.
The month it saves you
Not every month. Perhaps two or three a year, and in those it saves a wrong payment, a difficult conversation and a correction visible to everybody.
That is a reasonable return on twenty minutes twelve times.
Keeping the evidence
A note that it was done, and of anything found. It takes a line, and it is the difference between a business that checks and one that believes it checks.
Setting it up once
If the package does not provide the comparison, export two periods to a spreadsheet, put them side by side, and subtract. Saved as a template, it takes five minutes every month thereafter.
The reason this matters is that a control which requires reinvention each month is a control that will be skipped, and the reinvention is the expensive part rather than the checking.
What to do with the result
Keep it. A note of what was compared and what was found takes a line and turns a habit into something the business can say it does.
It is also what tells you, a year later, whether the check has ever caught anything — which is the only way to find out whether it is earning its twenty minutes.
The check against expectation, not against last month
One additional question worth asking at the same time: does the total match what you expected before you looked.
A comparison against last month catches changes you did not intend. A comparison against expectation catches a month where everything moved consistently for a reason nobody has noticed — a rate applied too broadly, a setting changed, a period keyed twice.
Whether the business can afford the month
Most small payrolls do not do this, and most are fine most of the time. The question is not whether it is necessary every month but whether the business can afford the month where it was.