$24 Million Recovered for Disability Support Workers — and the Regulator Mostly Went Looking Rather Than Waiting
$24 Million Recovered for Disability Support Workers — and the Regulator Mostly Went Looking Rather Than Waiting
Fifth in a series on the in-house versus fractional finance function. SCHADS award underpayment risk is rarely one bad pay run. It is the same small error repeating, which is exactly what a single-person payroll function is worst placed to see.
The Fair Work Ombudsman published its 2025-26 recoveries on 14 September. The headline was $453 million returned to more than 181,000 workers, a 27 per cent increase on the year before. Underneath it sat a figure that matters more to anyone running a disability or community services provider: $24 million recovered for 10,546 disability support workers, with 53 compliance notices accounting for $8.1 million and three enforceable undertakings signed.
The structural point is in how the money was found. In the disability sector, the FWO says the majority of recoveries came through proactive compliance activity rather than complaints. Across all sectors, proactive investigations rose 86 per cent, from about 1,400 to 2,680, recovering $45 million for 13,803 workers. That is a regulator increasingly arriving without being called.
|
$24m / 10,546
Recovered for disability support workers in 2025-26, across more than 1,000 completed disputes, 53 compliance notices ($8.1m) and three enforceable undertakings.
|
+86%
Increase in FWO proactive investigations across all sectors, to 2,680. In disability support, most recoveries came through proactive compliance work rather than complaints.
|
Divide by 10,546 and the shape of the problem appears
Roughly $2,280 per worker. An average cannot tell you how any individual case arose — the Ombudsman itself describes non-compliance as a continuum running from inadvertent mistakes to deliberate acts. But it is consistent with the pattern finance teams in this sector most often find when they do look: not a payroll catastrophe per person, but a rule applied slightly wrong, every fortnight, for a long time.
Which is the most useful thing to understand about SCHADS. The award's hard parts are not obscure; they are the ones that recur in every single pay run:
Classification and progression. A support worker sits in a level and pay point, and pay points advance with experience. In the common case nobody is deliberately underpaying; the progression date simply passes without anyone noticing, and then it never corrects itself.
Broken shifts. The allowance structure depends on how many unpaid breaks split the shift and how the span is calculated. Rostering systems handle the common case well and the second-break case less well.
Sleepovers. The rules changed from the first full pay period on or after 1 June 2026. Work immediately before and after a sleepover can now form a single shift of up to 12 ordinary hours by agreement, with no more than eight on either side, and shift loadings apply to each work period separately. The sleepover allowance, and any work performed during the night, are paid on top. A rostering system configured before June may still be applying the old logic.
Overtime and span. When a shift starts, when it finishes, whether the day of the week changes the rate, and whether time between shifts triggers a penalty.
Allowances. Travel between clients, kilometre rates, on-call, first aid, sleepover, and the ones that apply only in certain circumstances and so are the easiest to miss entirely.
Every one of these is a rule, and rules fail consistently — which is the good news, because a consistent failure is detectable, provided somebody is looking across pay runs rather than at one.
Why one person cannot see it, however good they are
This is not about competence. A payroll officer processing a fortnight for 90 support workers is solving a scheduling problem under a deadline: did everyone get paid, did the exceptions clear, is the file lodged. The work is transaction-level by design, and at transaction level a $21.81 one-break allowance paid where the $28.87 two-break allowance applied looks like nothing.
The error is only visible at pattern level — the same worker, the same shift type, the same variance, thirty times. And nobody sees pattern level while inside the pay run. You see it from outside, looking backwards, across periods.
Small finance teams almost never have someone outside the pay run. The person who processes payroll also reviews it, and self-review has a known ceiling: you cannot check your own interpretation of a rule, because you apply the same interpretation to the check. That is a design problem, not a performance one.
What a second set of eyes actually does
Pre-run validation. Before the run is committed, not after. A short set of checks that anyone can apply: any classification unchanged for more than 24 months; any worker whose average hourly rate has moved more than a few per cent without a known cause; any shift type appearing with no matching allowance; any employee paid below the relevant minimum for their classification once allowances are stripped out.
Anomaly review across periods. Take twelve pay runs and look at the distribution rather than the total. If 40 broken shifts produced 31 allowances, the 9 missing ones are the finding — and one of them will explain the other eight.
An interpretation log. Write down how you read each tricky clause, with the date and the source. When the FWO or an auditor asks, the difference between a documented interpretation applied consistently and an undocumented one applied inconsistently is very large, both in how it is treated and in how long it takes to resolve.
A re-check when the rules move. Which they keep doing. The sleepover rules changed in June, and the SCHADS Schedule E interim increase now applies from the first full pay period on or after 1 December, not 1 October — a date that moved after most FY27 budgets and plenty of secondary summaries were written. Every rule change resets the patterns above, and the first two or three pay runs after one are where new recurring errors are born.
Where AI fits, narrowly
Pattern detection across pay runs is genuinely well suited to automation: it is repetitive, rule-shaped, and the output is a list of exceptions for a human to judge. What automation must not do is decide whether a variance is correct. Award interpretation is a judgement with legal consequences, and it needs a person's name on it.
We are working on a payroll checking capability at PFL. It is at an early stage and I am not ready to talk about it in detail yet. What I will say is that the value is not in the checking itself; it is in having the check performed by something that was not also the thing that ran the payroll.
This post is general commentary based on publicly available information and does not constitute legal or tax advice. Award interpretation depends on your specific classifications, rosters and enterprise agreement. Always seek independent professional advice before acting.
On AI and your data: payroll files are among the most sensitive data an organisation holds. Before any of it reaches an AI tool, confirm in writing whether the vendor retains customer inputs for model training. The safest posture is a tool where your data is not retained for training at all — and, in line with OAIC guidance, keep names and identifiers out of anything the tool sees wherever the check allows. A no-training setting does not change your Privacy Act obligations.
Free Finance Function Health Check
A short, no-obligation review of how your finance function is structured, what it costs you fully loaded, and where the gaps are — including whether anyone is reviewing payroll from outside the pay run. PFL provides senior-level outsourced finance, management reporting, and AI automation for Australian NFP, NDIS, and SME organisations.
Book your free health check →Fair Work Ombudsman — FWO recovers $453 million for workers (14 September 2026)
Law Society Journal — Ombudsman recovers $453m in unpaid wages: what's behind the 'continuum of non-compliance'? (22 September 2026)
Fair Work Ombudsman — Pay and allowances in the SCHADS Award
Fair Work Ombudsman — Hours of work in the SCHADS Award (broken shifts, sleepovers, span)
Fair Work Ombudsman — Changes to sleepovers in the SCHADS Award (from 1 June 2026)
Fair Work Ombudsman — Social, Community, Home Care and Disability Services Industry Award [MA000100] summary
Office of the Australian Information Commissioner — Guidance on privacy and the use of commercially available AI products
Comments
Post a Comment