You already have a biometric device and a payroll package. What should an HRMS add?

A manufacturer told me they were well covered on HR technology: a biometric device at each gate and a payroll package their accountant used. Payroll still took three weeks, and one person spent most of the month assembling it.

Both tools worked exactly as designed. The device recorded punches. The package computed salaries. Everything between those two facts was being done by hand.

What sits in the gap

WHAT HAPPENS                          WHO HOLDS IT TODAY
---------------------------------------------------------------
Employee punches in                   biometric device
Which shift was he on?                supervisor's memory
Was a late arrival approved?          nobody / a message
Missed punch — who corrected it?      supervisor's word
Was leave approved, and by whom?      a chat thread
How much leave is left?               a spreadsheet formula
Was overtime sanctioned?              verbal
Is the month closed?                  no one is sure
Attendance summary to payroll         built by hand
Salary computed                       payroll package

Two tools at the top and bottom. Nine lines in between held
by one person, a spreadsheet and several conversations.

That middle block is not a software gap in the abstract. It is the specific reason payroll is late, why overtime is disputed, and why nobody can answer a leave-balance question without checking.

What a device genuinely cannot do

• Know the shift. A punch at 21:40 is either a late arrival, an early night shift or an overtime start. Only a roster tells you which.
• Hold an approval. A device records that a punch is missing; it has no view on whether the absence was sanctioned.
• Carry a correction with a name on it. The correction happens elsewhere, usually verbally, and leaves no trace.
• Compute a balance. Accrual, carry-forward and consumption are rules, not readings.
• Close a month. There is no cut-off in a device, so corrections keep arriving after payroll has started.

None of that is a criticism of the device. It is a record of presence, and it does that well. The mistake is treating it as an attendance system.

What an HRMS adds, concretely

1. 1

   The roster, so a punch means something

   Shift patterns, rotations, weekly offs and who was scheduled. Without this, night shifts crossing midnight land on the wrong day and nobody can explain the month.

2. 2

   Approvals with a trace

   Leave, regularisation and overtime approved by a named person, against a date, retrievable a year later. This is the single largest improvement for most businesses.

3. 3

   A balance the employee can see

   Which removes most of what reaches HR in a week, and removes the argument about arithmetic neither side can check.

4. 4

   A cut-off that actually closes

   After which corrections become next month's arrears rather than a reason to recompute. A device cannot enforce this; a system can.

5. 5

   The handover to payroll

   A frozen, approved input set produced on a known date, instead of a spreadsheet assembled from three sources. That is where the three weeks go.

Together those are what attendance, shift and leave software is for — and none of them requires replacing either the device or the payroll package.

You probably keep both tools

This is worth saying plainly, because businesses often assume an HRMS means starting again. Most devices export punches in a form a system can read, and most payroll packages accept a structured input file. The device stays at the gate and the package stays with your accountant.

What changes is the nine lines in the middle. That framing also makes the buying decision much narrower: you are not evaluating a payroll product, you are evaluating whether something can hold your roster rules, your approval routes and your cut-off. Which is a far more answerable question, and the one the payroll input management side of a system exists to answer.

The test to run in a demo

Take one real week from one department — including a night shift, a missed punch, an approved leave and a sanctioned overtime — and ask the vendor to process it. Not sample data. That week.

Watch specifically for where the night shift lands, how the missed punch is corrected and by whom, and what happens when you try to submit a correction after the cut-off. Those three behaviours separate products more reliably than any feature list.

And the part software will not fix

If your cut-off is a suggestion, if missed punches are fixed on a supervisor's word, and if nobody owns the monthly close, a system will enforce rules that do not currently exist — and it will be experienced as rigid rather than helpful.

So the rules come first, even if only on paper for one month. Published cut-off, named approver per correction, a regularisation window that closes before payroll opens. That is attendance, shift and leave process work rather than a purchase, and businesses that do it discover how much of the three weeks was process rather than tooling. If you want to see the software half run against your own week instead of a demo script, book a free demo of GullyHR software.

A worked comparison

Take a single employee on a rotating shift who misses a punch on a Tuesday night. With a device and a payroll package alone, the supervisor mentions it, somebody notes it on a sheet, and payroll accepts it. Three facts are missing: which shift he was rostered on, who authorised the correction, and whether it was raised before the cut-off.

With the middle layer in place, the same event produces a record: rostered night shift, correction requested by the employee, approved by the named reporting manager on the Thursday, inside the window. The salary is identical. What differs is that in March somebody can answer the question.

The question to settle before buying anything

Businesses in this position — a device at one end, a payroll package at the other — frequently describe the gap as a technology problem. It is worth testing that before spending anything.

The test is to write down what happens between the device and the payroll run, step by step, naming who does each step. Most businesses have never done this, and the exercise takes about twenty minutes. What it typically reveals is that the manual work is not extraction, which is quick, but adjudication: someone deciding what to do about punches that do not make sense.

That distinction determines what to buy. If the effort is extraction and reformatting, the problem is integration and it is straightforward. If the effort is adjudication, no amount of integration removes it, because somebody still has to decide — and what helps is a system that routes those decisions to the right person with the context attached, which is a different capability.

Buying for the first problem when you have the second is the most common disappointment here. The data flows cleanly, the monthly effort barely moves, and the conclusion drawn is that the software did not work when the requirement was never correctly stated.

Run the twenty-minute exercise before any demo is booked, and take the written steps into the demo itself. A vendor shown a specific sequence will either demonstrate how their product handles it or will change the subject, and both responses are informative in a way that a general capability walkthrough is not.

Multiply that by a few hundred corrections a year and it stops being an administrative nicety. It is the difference between a month you can defend and one you cannot, which is what attendance, shift and leave software is actually buying you.