Underpayment Usually Enters at the Keyboard
Most Australian underpayment is not a decision. It is a transcription.
Watch where it happens. A manager writes up the fortnight. Someone, often that same manager late on a Sunday, or a bookkeeper working across four clients, types those hours into payroll. At that keystroke every award judgement gets made at once: which hours are ordinary, which attract a Saturday rate, whether that Sunday shift crossed midnight, whether the split counts as a broken shift, whether the casual loading sits on top or is folded in.
Get one of those wrong and it is wrong for that person for that fortnight. Get it wrong because you misunderstood the rule rather than mistyped it, and it is wrong for everyone, every fortnight, until somebody notices. That second version is what turns into a remediation program, and it started as an ordinary person doing data entry without an award open next to them.
Integration does not make you compliant. What it does is remove the place where the error is introduced, by carrying the categorisation decided at roster-build time through to the pay run rather than asking someone to re-derive it from a timesheet.
How the Flow Actually Works
The mechanics are similar across platforms.
Connect, or set up the export. With a direct integration such as Xero Payroll you log in to the payroll platform and grant access; credentials are not copied between systems and permission can be revoked from the payroll side at any time. With a CSV export, as used for MYOB and QuickBooks, there is no live connection: you generate a file and import it, which means someone owns that step every pay period.
Match the employees. The two systems have to agree on who is who. This is the step that breaks first exports, almost always because a name differs, a nickname is used in one system, or someone exists in payroll and not in rostering. Fixing it is tedious rather than hard, and it is much cheaper before the first run than during it.
Build the roster with rates applied. Penalty rates attach at build time based on day, time, employment type and classification, which is also what lets a manager see the cost before publishing. Note that shift cost breakdowns are a paid feature in GetMyRoster rather than part of the free plan.
Capture what actually happened. Rostered hours and worked hours are not the same thing, and paying from the roster means paying from an intention. A GPS time clock, also a paid feature, records actual start and finish times.
Approve, then export. Approval is a real control, not a formality, and it should sit with someone who would notice a shift that looks wrong. The export then sends hours split by category into the payroll platform for the pay run.
Xero, MYOB and QuickBooks
| Platform | How it connects | Set up before your first export |
|---|---|---|
| Xero Payroll | Direct API integration | Earnings rates for each penalty category; employees active in payroll |
| MYOB | CSV timesheet export | Payroll categories matching your award loadings |
| QuickBooks | CSV timesheet export | Payroll items for each rate type; employee time records enabled |
The pattern across all three is the same and it is the thing to take away: the payroll platform needs a category to receive each kind of hour before the export will land correctly. That applies equally to a direct integration and to a CSV import. If Sunday hours arrive and there is no Sunday earnings rate configured, they do not fail loudly. They land somewhere plausible, usually ordinary time, and get paid at the wrong rate. Set the categories up first, then export.
Integration specifics change as vendors update their APIs, so confirm current capability for your exact payroll product and version rather than relying on any article, including this one. Note also that payroll integration sits on the GetMyRoster Enterprise tier rather than the free or per-user paid plan.
What Integration Does Not Do
Worth being blunt, because this gets oversold.
It does not run your payroll. Tax, superannuation and Single Touch Payroll reporting remain the payroll platform's job. Rostering feeds payroll; it does not replace it.
It does not interpret your award. It applies the rates you configured. Configure them wrong and you get consistent, confident, automated errors, which are harder to spot than manual ones precisely because they look systematic. Have someone who knows the award check the configuration before the first live run, and again after any annual wage review.
It does not transfer liability. Under the Fair Work Act 2009 the obligation to pay correctly stays with the employer regardless of which systems produced the number.
It does not fix bad source data. An employee recorded under the wrong classification at onboarding will be paid wrongly with complete consistency, forever, until a human notices. Integration accelerates whatever you feed it.
The honest framing is that integration converts a recurring manual risk into a one-off configuration risk. That is a very good trade, because configuration gets checked once and carefully, whereas data entry happens every fortnight at speed by whoever is available. It is not the same thing as compliance.
Getting the First Run Right
Reconcile one full cycle in parallel. Run the old process alongside the new for one complete pay period and compare line by line. Not a spot check. This is the only step that reliably surfaces miscategorised hours, and almost nobody does it.
Test your awkward cases deliberately. A shift crossing midnight. A public holiday. A broken shift. A casual on a Sunday. Under SCHADS that last one should come out at 225%, being the 200% Sunday rate with the 25% casual loading on top, and if your configuration returns 200% you have found a real problem before it cost you anything.
Check the leave interaction. Confirm how approved leave appears in payroll and whether accruals are affected, because leave handling differs between payroll products and employment types.
Re-check after every wage review. The Fair Work Commission adjusts award wages annually, usually effective from the first full pay period on or after 1 July. Rates configured last year are wrong this year, and nothing in the software will tell you.
Keep the records. Time and wages records must be kept for seven years. Under section 557C, an employer without adequate records carries the burden of disproving an underpayment allegation, which is covered further in Fair Work rostering requirements. Providers in regulated sectors should also read NDIS rostering.
The Failure Modes Worth Knowing About
Four things go wrong with roster-to-payroll integrations, and they are the same four almost every time.
The silent miscategorisation. Covered above and worth repeating because it is the expensive one. Hours arrive in payroll with no matching category and get absorbed into ordinary time. Nothing errors. The pay run completes. Everyone is underpaid by exactly the penalty component, every fortnight, until someone reconciles. Guard against it by exporting a test period first and checking that the category totals in payroll match the category totals in the roster, rather than only checking that the grand total looks plausible.
The employee who exists twice. Someone is created in payroll while also being created in rostering, the match is made on name, and a spelling difference produces two records. Hours split across both and neither looks obviously wrong. Audit the employee mapping whenever anyone joins, and treat payroll as the system of record for identity.
The unapproved timesheet that exported anyway. If approval is configured as optional, it will eventually be skipped during a busy fortnight and unreviewed hours will reach the pay run. Make approval mandatory and make the approver someone who knows what a normal week looks like for that team.
The July rates nobody updated. Annual wage review changes usually take effect from the first full pay period on or after 1 July. Configured rates do not update themselves, and the software cannot know they are stale. Put a recurring calendar reminder in June, check it against the current award, and confirm the first July pay run before it is finalised rather than after.
Running through all four is the same underlying point. Integration is excellent at doing consistently whatever it was told to do, which means the quality of the configuration and the quality of the review process determine the quality of the outcome. The systems that get into trouble are not the ones with no automation. They are the ones where automation replaced attention entirely, and nobody reconciled anything for two years because the exports kept succeeding.
Frequently Asked Questions
Connect Your Roster to Payroll
Export award-categorised hours to Xero Payroll directly, or to MYOB and QuickBooks by CSV. Start free for up to 5 users at one location; payroll integration is an Enterprise feature.
Start Free - No Credit Card