Who this is for. HR holds the employee. Rostering plans the shift. Time and attendance records what happened. Payroll pays for it. Between the systems that plan and record work and the system that pays for it sits a step none of them owns: interpreting the award or agreement — deciding which penalties, allowances, loadings and entitlements the hours actually worked have triggered. In most organisations that step has no home in the stack, so it is done by a person, in a spreadsheet, each cycle — and it is where underpayment, rework and lost evidence originate. This note makes the case for filling that gap with a control that runs before payroll processes — alongside the systems you already run.
The stack, and the gap in the middle
Most organisations already run several systems across the pay process. Each does its job. None of them interprets the award.
- HR knows who. The employee, their classification, the award or agreement that covers them, their standing entitlements. It does not know what a given fortnight's hours should pay.
- Rostering knows the plan. The shifts intended. It does not know the entitlements those shifts will trigger once they are actually worked.
- Time & attendance knows what happened. The actual hours — increasingly captured tamper-proof through facial recognition. It records the hours; it does not interpret them.
- Payroll applies configured rules and pays. The edge cases it cannot model — the conditional allowance, the unusual penalty interaction, the agreement clause that varies the award — come back marked "calculate manually."
Interpreting the award against the hours actually worked — the step that turns a record of attendance into the correct entitlements — sits between these systems, owned by none. So it happens where nothing else will do it: by hand, in a spreadsheet, every cycle.
The interpretation step is not missing from the stack. It is simply happening in a spreadsheet — the one place in the pay process with no controls, no coverage guarantee, and no evidence trail.
The rules that don't fit a field
Award interpretation resists configuration because the rules are conditional and interacting — logic, not values in a dropdown. The recurring hard cases:
- Conditional allowances. Paid only when a condition is met — a location attended, a task performed, a qualification used, a piece of equipment operated. The trigger is in the work, not the pay code.
- Penalty and overtime interaction. Which base overtime compounds on; whether penalties and overtime stack; where the threshold falls in a shift that crossed midnight.
- Span, breaks and engagements. Ordinary-hours windows, minimum breaks between shifts, broken-shift rules, minimum engagement periods.
- Accruals and banking. RDOs, time off in lieu, service-based entitlements — accrued, banked and drawn down across periods rather than within one.
- Annualised and set-off reconciliation. Proving a salary or over-award payment still absorbs the award entitlements it is meant to cover, within the period the award requires.
- Agreement overlays. An enterprise agreement that varies the award in places — and the interaction between the two is exactly where a configuration field has nothing to say.
None of these is exotic. Each is routine. Together they are why the spreadsheet exists.
Interpretation belongs before the pay run
Where interpretation happens determines what it costs.
| When the award is interpreted | What a discrepancy becomes |
|---|---|
| Before payroll processes | A fix — corrected before anyone is paid. |
| After the run (complaint, audit, review) | A correction — with back-pay, interest and reputational cost. |
| Never, independently | A finding — waiting to be made by someone else. |
The cheapest place to interpret the award is before the money moves. That makes interpretation a pre-payroll step by definition — not a downstream reconciliation.
Making interpretation a control
A spreadsheet is interpretation without control. The same logic, encoded as a controlled step, changes its character entirely:
- It runs every cycle, for every worker — not a sample when someone has time.
- It is independent of payroll's configuration — so it can contradict payroll, which a check that shares payroll's rules never can.
- Its rules are versioned and citable — every result names the clause and the rule version behind it.
- It retains evidence automatically — the hours, the rule, the result, per worker, per period.
- It runs before export — surfacing discrepancies with the expected value attached, in time to fix rather than correct.
Where Smartta fits
Smartta fills that gap. It reads from the systems you already run — the roster, the attendance record, the HR classifications — interprets the award and agreement against the hours actually worked, including the rules your WFM and payroll leave manual, validates the result before the pay run, and hands the computed, compliant pay to payroll with the clause and the record attached.
It replaces none of them. HR stays your system of record; rostering stays your planning; time and attendance stays your capture; payroll stays your payment. Smartta is the interpretation and control layer between them — the step filled today by a spreadsheet, made a control that runs, covers everyone, and leaves evidence.
What to require
- Interpretation is encoded, not manual. The rules your payroll flags "calculate manually" are modelled — not left to a spreadsheet.
- It runs before payroll processes. Discrepancies are surfaced in time to fix, not correct.
- Coverage is complete, not sampled. Every worker, every cycle.
- Rules are versioned and citable. Every figure names the clause and rule version behind it.
- Validation is independent of payroll. A check that shares payroll's configuration reproduces payroll's assumptions.
- Evidence is retained automatically. Hours, rule, result — per worker, per period, without reconstruction.
Prepared as general guidance on payroll compliance. Not legal advice; the application of an award or agreement depends on the specific instrument and circumstances.