Accounting Staff Shortage Australia, AI Is Not the Enemy

Last reviewed: September 2026

Is there actually an accounting staff shortage in Australia?

Yes, but not the one most articles describe, and the difference changes what you should do about it. The official Australian source is the Occupation Shortage List published by Jobs and Skills Australia. The 2025 list, released in October 2025, rated 293 of 1,022 assessed occupations in national shortage, or 29% on the OSCA 2024 classification basis, down from 33% in 2024 and a 36% peak in 2023. The Australian labour market as a whole is easing.

Accounting is not easing evenly. On the 2025 list, Taxation Accountant and External Auditor are rated in shortage nationally and in every single state and territory. Accountant (General), Management Accountant, Internal Auditor, Bookkeeper, Accounts Clerk and Payroll Officer are all rated no shortage. If you only read the headline, you conclude that Australia is short of accountants. If you read the list, you find something more useful: Australia is short of the licensed, judgement-heavy roles, and is not officially short of the processing roles.

The professional bodies' own data points the same way while disagreeing on the margins. CA ANZ surveyed 159 members who had advertised vacancies in Australia between January and December 2025 and submitted the results to Jobs and Skills Australia's 2026 consultation. As reported by Accountants Daily on 7 April 2026, fill rates came in below the 67% threshold that signals a likely shortage for four roles, and CA ANZ recommended general accountants, taxation accountants and external auditors be added to the 2026 list. That list is due in October 2026.

Why can I not hire a bookkeeper if bookkeepers are not in shortage?

Because the occupation is not what you are hiring for. You are hiring for a person who can pick up a client file in Xero or MYOB, understand Australian GST coding, deal with a payroll that has three award interpretations in it, and not need checking twice. CA ANZ recorded the reason its members' vacancies went unfilled: a lack of experienced professionals, with respondents also naming a dearth of specialised skills in areas such as tax. Applicants exist. Applicants who can run a file unsupervised do not.

This is why the average practice ends up in the same shape. The senior tax accountant, the one the 2025 Occupation Shortage List says is genuinely scarce, spends part of every week chasing bank statements and recoding a client's expenses. Not because that is a good use of them, but because the alternative is that it does not get done. That is the actual failure mode of the Australian accounting shortage in 2026: the scarce resource absorbs the abundant work, and the practice loses the margin on both.

What about hiring offshore or sponsoring someone from overseas?

Both are legitimate and plenty of Australian firms do them. Neither is fast. Sponsored migration under the Skills in Demand visa carries a salary floor before you count anything else: the Core Skills Income Threshold increased by 3.8% to AUD 79,499 for nomination applications lodged from 1 July 2026, up from AUD 76,515, indexed to Average Weekly Ordinary Time Earnings, as set out by Baker McKenzie on 2 March 2026. That is a floor, not a salary: the annual market salary rate for the role and location applies on top, and the occupation has to be on the relevant list for the Core Skills stream before any of it matters. Add sponsorship, nomination and visa processing, and a four-partner practice is committing to a multi-month, five-figure process for one seat.

Offshore engagement is quicker but it is still hiring. You are adding a person who needs onboarding, review capacity, a security arrangement over client data, and someone in Australia to manage them. Practices that treat offshoring as a cost saving rather than a capacity build usually discover the review burden lands back on the same senior who was already the bottleneck.

The global picture is instructive, though it is not Australian-specific. Advancetrack's 2026 Accounting Talent Index, a survey of firm leaders across the UK, US, Canada and Australia by a firm that sells outsourcing, reports that 73% of respondents say talent shortages are having a severe impact, 73% are turning away potential clients, and 45% say the shortage is worse than three years ago. On responses: 72% are exploring outsourcing, offshoring or technology to manage capacity, and 16% are actively investing in AI. Read that gap carefully. The overwhelming majority are looking for capacity anywhere they can find it, and only a sixth have moved on the option that does not require recruiting a human being.

Will AI take accounting jobs, or fill the ones you cannot?

In Australian practices right now, overwhelmingly the second. The forum argument treats AI adoption as a substitution decision, as though a partner sits down, compares an AI to a graduate, and lets the graduate go. That is not the pattern Agentive sees. The pattern is a practice with an unfilled seat, a waiting list of prospective clients, and a senior who is at capacity. The AI goes on the work nobody is currently doing, or the work the wrong person is currently doing.

This is Agentive's first-party observation from deploying dedicated AI into Australian finance teams, and it is the single most reliable predictor of whether a rollout works: practices that buy an AI to take on a client they would otherwise have declined get a result they can measure within a quarter. Practices that buy one to reduce a wages line rarely do, because the wages line does not move until the work actually stops arriving on that person's desk, and it never does by itself. We have written the broader career question up separately in will AI replace accountants and bookkeepers, and the head-to-head hiring comparison in choosing between an AI and a human for the same role.

There is one honest counterweight and it deserves stating plainly. If practices automate the entry-level processing work, they remove the training ground that produces the senior accountants who are already scarce. The 2025 Occupation Shortage List problem gets worse in five years, not better, unless firms deliberately rebuild how juniors learn. That is a real risk to the profession. It is not the same as AI taking the job of the person currently doing it.

Which work should a practice hand to a dedicated AI first?

Prioritise work that is high volume, low in judgement, and currently being done by your most expensive person because nobody else is available. That third condition is the one firms skip, and it is where the return actually comes from. Automating a low-value task that a junior does cheaply saves you very little. Automating the same task when your registered tax agent is doing it at 7pm converts scarce, licensed capacity back into billable work.

The mechanics of the first three rows are covered in more depth on Agentive's use case page for bookkeepers and in the bookkeeping capability overview. The reason they sit at the top of the list is that an error in any of them is visible immediately, which is exactly the property that makes work safe to automate.

What does the capacity maths actually look like?

Here is a worked example, labelled as such rather than presented as a client result. The figures are assumptions you should replace with your own.

1. Establish the leak. A four-person suburban practice finds its registered tax agent spends about nine hours a week on document chasing, coding and workpaper assembly, because the seat that used to do it has been empty since March.
2. Price the hiring option honestly. Using the CA ANZ figures, a taxation accountant vacancy took an average of 77 days to fill among firms that filled it, and only 55% of advertised vacancies were filled at all. The expected outcome of advertising is not "a person in 77 days". It is roughly even odds of a person in about three months.
3. Price the migration option. A sponsored Core Skills nomination lodged after 1 July 2026 must pay at least AUD 79,499, plus the market salary rate, plus sponsorship, nomination and visa costs, plus processing time.
4. Set a realistic absorption rate. Assume a dedicated AI takes six of the nine hours and escalates the remaining three as exceptions. That is about 270 hours of senior capacity a year, on a 45-week basis.
5. Decide where the hours go before you start. If those 270 hours are re-routed into advisory or into onboarding the client the practice declined last quarter, the capacity has a price. If they are absorbed into finishing earlier, the practice has bought a nicer evening, which is worth something but is not what was on the business case.

Step five is the one that decides the outcome, and it is not a technology step. Reclaimed hours do not convert themselves into revenue. A practice that cannot name, in advance, the specific client work that will fill the gap should expect the gap to fill itself with more of the same work. We have gone deeper into how this changes the fee conversation in what to charge when AI does the compliance.

What does not work, and where do firms waste the money?

AI does not fix a broken workflow. It runs a broken workflow faster. If client documents arrive three weeks late, the chart of accounts has four variations of the same expense category and no one owns the file, adding an AI produces the same problems at higher volume and with less time to notice them. Fix the intake process and the coding rules first. That work is unglamorous and it is the actual prerequisite.

Three other failure modes are worth naming. Firms buy a general-purpose chat tool and expect ledger accuracy, which produces the traceability problem covered in our post on hallucinated figures in financial reports. Firms skip the integration question and end up with a tool that cannot read the file it is supposed to work on, which is why the connection layer into Xero and QuickBooks matters more than model choice. And firms treat the AI as an unsupervised staff member, which is a compliance problem rather than a technology one.

On that last point, Australian tax and BAS agents have explicit guidance. TPB(GS) 55/2026, issued 22 July 2026, states that tax practitioners remain accountable for the accuracy of information and advice they provide to their clients, and that AI models may hallucinate or provide inaccurate information and cannot be relied on as a replacement for tax knowledge, experience or expertise. Capacity gained by removing the review step is not capacity, it is exposure. The disclosure and consent side is set out in our companion piece on what tax and BAS agents must tell clients.

What should an Australian practice do in the next month?

Four steps, in order, and none of them start with a vendor demonstration.

1. Count what you turned away. List the prospective clients your practice declined or did not chase in the last twelve months, with the fee you would have charged. That number is the size of the problem, and most practices have never written it down.
2. Find the misallocated hours. Take one week of timesheets and mark every hour your most senior people spent on work that does not require their registration. Score each task on volume and judgement using the table above.
3. Fix the intake before automating it. Standardise how source documents arrive and clean up the chart of accounts for your three messiest clients. If you cannot do this, automation will amplify the mess rather than resolve it.
4. Pick one task, one client, one month. Run it with a review checkpoint and an exception queue, and measure the reclaimed hours against your own timesheets rather than a vendor's estimate.

Agentive builds for exactly this shape of problem. Agentive's AI Employee for accountants runs single-tenant on AWS Sydney with all inference inside Australia, keeps a per-action audit log so every step is reviewable, and escalates rather than guessing when a document is missing. Client data never leaves Australian borders and is never used to train a model, which is the baseline any Australian practice should insist on before a client file goes near a tool.

The forums will keep arguing about whether AI takes accounting jobs. The commercially useful question for an Australian practice in 2026 is narrower and more answerable: what is the work you already cannot staff, who is doing it now, and what is it costing you that they are. Jobs and Skills Australia says taxation accountants and external auditors are scarce in every state and territory. CA ANZ members filled barely half the vacancies they advertised. Against that, an AI Operation Engine is not a competitor to your team. It is the only capacity you can add this month.