History Repeating? What providers need to understand about Functional Capacity Assessments

The independent assessments story

Part three of a seven-part series on the 22 April 2026 NDIS reform announcement and what it is asking of providers.

On 22 April 2026, Minister Butler used his National Press Club address to set out a substantial reform agenda for the NDIS. The headline figure was $15 billion in projected annual savings by 2030. The agenda covered a raft of changes, but one of the foundational pieces is the introduction of functional capacity assessments, including the reassessment of all current participants and a "repositioning" between 160,000 to 300,000 to other currently undefined support programs.

The independent assessment proposition is the longest-running of the reforms in Minister Butler's announcement. It has been put forward in different forms by ministers from both sides of politics, withdrawn under state-government pressure, restated, modified, and is now being announced again. The rollout has also already shifted, formally pushed from mid-2026 to no earlier than 1 April 2027 in a further announcement on 29 April 2026.

This part of the series sits the 2021 proposal next to the 2026 proposal. It looks at what is identical, what was specified in 2021 and is not yet specified in 2026, and the conditions that may make this version land where the previous one did not.

It is not an argument for or against the proposal. It is an argument that the proposal needs to be assessed on its current design, rather than on the rhetoric of the announcement or on the assumed equivalence to a previous version that was rejected by states.

The emotional weight this announcement is carrying

Before any of the technical analysis, we need to acknowledge the emotional toll this announcement likely had (and is still having) on the people inside it.

Participants and self-advocates who fought in 2021 will be experiencing 2026 as the same thing returning in better clothes. The reaction will likely be visceral distrust, fatigue, and a sense that organised resistance from last time was absorbed and rebranded.

Families and informal carers will feel a sharper fear. The fear is that an assessment conducted on a bad day, in a clinical environment, by a stranger, will capture a performance of function rather than the reality of need, and that the plan will be sized to the performance.

Providers will be sitting with a different discomfort. Provider input is likely to be sought as evidentiary material in assessments, while the same providers know their revenue is shaped by the outcome of those assessments. That is a structural conflict of interest that the design has not yet addressed.

None of this is reason to dismiss the proposition. It is reason to be careful with how it is communicated, and honest about what it is asking the people inside the scheme to carry while the design is still being set.

Why functional assessment is needed, and why it is hard

You cannot fund support needs without defining what those needs are. The difficulty is that disability presents as broadly as the general human condition, and two people with the same formal diagnosis can have very different support needs.

Replacing diagnosis-based access with functional capacity assessment removes a structurally perverse incentive. Under the current lists, people have had to maximise diagnostic framing to secure access. That has driven adversarial dynamics with the agency and inflated the cost of diagnostic documentation. It also forces us to then rely on the deficit based documentation when designing daily supports.

The proposition is responding to a real design problem.

But the harder question is whether any instrument can do what the proposition is asking it to do.

From my research (and I would love to know if I'm wrong) there is currently no validated instrument, in any jurisdiction, that has operationalised a clean separation between intrinsic functional need, environmental disablement, and system-induced capability suppression. The three are deeply entangled in real life, and participants do not experience their need as decomposable into three columns. Most participants will struggle to articulate their needs in a form that maps cleanly onto items in a tool.

Even if such an instrument existed, the surrounding data architecture would systematically miss the clinical labour that keeps complex participants stable. Skilled prevention is invisible in incident data, because the whole point of skilled prevention is that the incidents do not happen.

John McKnight's argument was that professionalised systems produce dependency by replacing capability with service. There is a measurement-driven version of the same problem inside this proposition. When skilled prevention is working, the system reads the absence of incidents as the absence of need, withdraws the support, and creates the very incidents the support was preventing.

In my opinion, the two preconditions for a defensible national functional-capacity gate are currently missing. There is no discriminant instrument, there is no prevention-effort signal and the 22 April announcement does not name how either will be resolved before rollout.

What is identical between 2021 and 2026

Both the 2020-21 proposal and the 2026 proposal replace diagnosis-based access (Lists A and B under the NDIS Act) with a standardised functional capacity assessment. Both apply the test at access and at reassessment. Both apply across the whole scheme. Both feed the eligibility decision. The architecture is the same.

This is not coincidence. The proposition responds to a real and persistent design feature of the scheme. Diagnosis-based access produces inequities for people whose disability is episodic, whose function varies materially from day to day, whose diagnosis sits outside the canonical Lists A and B, or whose needs are functional rather than categorical. The Productivity Commission flagged this in 2011. Successive reviews have flagged it since. The structural critique that motivates the proposition has not changed.

What has changed, between 2021 and 2026, is the political and legislative environment around the proposition. That matters more than the architecture.

What we know now, side by side

The comparison below sets out what each version of the proposition specified at an equivalent stage of public communication. Items marked "not yet specified" are not necessarily a weakness in the 2026 design, they are the live questions that will be resolved in subordinate rules and co-design.

Core architecture 2020-21: Standardised functional capacity assessment replacing Lists A and B. 2026: Same.

Application points 2020-21: Access and reassessment, scheme-wide. 2026: Same.

Cohort affected 2020-21: All current and prospective participants. 2026: Same, with ~160,000 nominated for "repositioning" out of ~760,000.

Assessment tools 2020-21: Six named instruments (incl. WHODAS 2.0, Vineland-3, PEDI-CAT). 2026: Not yet named publicly. I-CAN v6 signalled in agency materials.

Assessor panel 2020-21: Eight contracted panel organisations named. 2026: Not yet named. Accreditation framework signalled.

Role of treating clinicians 2020-21: Contracted assessors only, treating clinicians excluded from the decision. 2026: Not yet specified. Implementation imperatives point to a contracted-assessor model.

Workforce model 2020-21: Published. 2026: Not yet published.

Procedural pathway 2020-21: Referral-to-decision steps set out. 2026: Deferred to subordinate rules and co-design.

Technical Advisory Group 2020-21: In place. 2026: Not constituted as at the speech date.

Budget translation engine 2020-21: Not publicly released; tool-to-budget mapping was opaque. 2026: NDIA designing its own IT system (POSIT / "budget model engine") to convert I-CAN scores to dollar amounts. Confirmed at Senate Estimates, December 2025.

Co-design sequencing 2020-21: After tool and panel selection. 2026: Before tool selection (signalled).

Procedural additions 2020-21: None named beyond the core process. 2026: Accreditation, phased rollout, emotional safety testing, and a personal and environmental circumstances questionnaire.

Legal authorisation pathway 2020-21: Required state agreement through the NDIS architecture. 2026: Subordinate rules under the Getting the NDIS Back on Track Act 2024.

State veto power 2020-21: Effective; states blocked the proposal. 2026: Materially reduced; communique-level objection no longer enough to stop a Commonwealth pathway.

Sector position 2020-21: Disability peaks in a "scrap it" coalition. 2026: DANA-led joint statement of 22 April 2026, signed by twelve organisations, in a "build it with us" position.

Rollout timing 2020-21: Announced for mid-2021, withdrawn. 2026: Announced for mid-2026, formally pushed to no earlier than 1 April 2027 (29 April 2026).

Review rights 2020-21: Thin pathway through the AAT for plan decisions. 2026: Through the ART. The ART can order the NDIA to do a new Support Needs Assessment but cannot substitute its own finding on funding quantum.

What 2021 had that 2026 has not yet specified

The 2020-21 proposal was technically more advanced than the 2026 proposal at the equivalent stage of public communication. It named the tools, the panel, the workforce model, and the procedural pathway. The detail was on the table when public engagement began.

Butler's 22 April announcement names no tools, no workforce model, no role for treating clinicians, and has no Technical Advisory Group constituted as of the speech date. The detail has been deferred to subordinate rules and a co-design process to be run by the agency.

This is not necessarily a weakness in the 2026 proposal. It may be a deliberate response to one of the lessons of 2021, where an already-procured tool list and panel of assessors meant that disability peaks and state ministers were arguing against a fait accompli rather than shaping a design in formation. Co-design before tool selection is a different sequencing than tool selection before co-design.

A more critical gap sits underneath the tool question. The budget translation mechanism, which converts assessment scores into plan dollars, has not been publicly released. Senate Estimates in December 2025 confirmed the agency is designing its own IT system for this translation. Most public commentary is focused on which tool will be selected. The tool sets the score. The translation engine sets the budget. The assessment tool and the funding algorithm are two totally separate parts that need to merge to make this successful and this distinction was largely missed in 2021. Currently, it remains the part of the 2026 design that is least visible in the public conversation.

The risk in the deferred-detail approach is that the operational consequences of the proposition cannot be tested against its design until that design exists. Providers and participants are being asked to engage with a proposition whose substance will be set in subordinate rules that do not yet exist. The Senate, in its scrutiny of subordinate legislation, has limited capacity to disallow rules made under enabling legislation that has been drafted broadly. Many of the structural questions about the proposition will be resolved at a level of legislative scrutiny that is materially less than primary legislation.

Why this version might land where 2021 did not

The 2026 design includes procedural additions that the 2021 model did not have. Accreditation of assessors, a phased rollout, emotional safety testing of the experience, and a personal and environmental circumstances questionnaire alongside the core instrument are all signalled. None of those by themselves resolve the validity questions, but they do change the lived experience of the assessment. They also give the design the appearance of having absorbed the 2021 critique.

Additionally, the 2026 proposal sits on legislative architecture that did not exist in 2021. The Getting the NDIS Back on Track Act 2024 created enabling powers that allow the Commonwealth to push detail into subordinate rules. The 2021 proposal had to negotiate the substance through state ministers because the scheme architecture required state agreement. The 2026 proposal has more legal capacity to proceed without it.

That is a different starting position. State ministers in 2021 were able to block the proposal because of the operational architecture of the scheme. State ministers in 2026 may still object, but their capacity to prevent commencement is materially reduced. I'm hypothesising that in the current environment, state based objection will not be enough to stop a Commonwealth-driven implementation pathway that has been authorised by primary legislation.

But, that is assuming there is an objection to be had. The sector consensus has also shifted. In 2021, disability peaks were in a "scrap it" position. The DANA-led joint statement of 22 April 2026, signed by twelve organisations, has moved to a "build it with us" position. The themes are similar (threshold setting, receiving services first, co-design substance) but the intent is constructive rather than oppositional.

The 2021 concerns that have not moved

Several concerns from 2021 remain unanswered in the 2026 announcement.

The first is the use of contracted assessors rather than treating clinicians. Functional capacity tests delivered by an assessor who has not seen the participant before, in a one-off appointment, against a standardised tool, will produce systematically different results to functional capacity tests delivered by a treating clinician who has the longitudinal record. The 2021 model used contracted assessors. The 2026 model has not yet specified, but the precedent and the implementation imperatives both point to a contracted-assessor model. This then leads into a critical structural problem, already rife in the scheme. Role definition. The Assessor role will, by virtue of the outcome sought, contain a built-in identity conflict between a therapeutic orientation and a gatekeeping function, and that conflict is likely to compress scores downward under medico-legal pressure. We have seen the same dynamic in the variability of NDIS-accredited auditors, where accreditation has not produced consistent practice. Inter-rater reliability of FIM-style and ICF-based instruments is also documented to deteriorate outside controlled trials, in aged care and workers' compensation settings. That is not an argument against the proposition. It is an argument that assessor variability needs to be designed for, not assumed away.

The second is the use of standardised tools that may not capture intersectional or episodic disability. A standardised tool produces consistent results for the conditions it was designed to assess. People whose disability sits across multiple intersecting conditions, or whose function varies materially across days or episodes, are systematically under-captured by single-instrument standardised assessment. The 2021 tools had this issue. The 2026 toolset has not been specified, but the structural question remains.

The third is procedural fairness and review rights. The 2021 proposal had a thin pathway for participants to challenge an assessment outcome. The Administrative Review Tribunal (the successor to the AAT) does have review rights over new framework plans, but with a critical structural limitation. The ART can only order the NDIA to do a new Support Needs Assessment. It cannot substitute its own finding on funding quantum. That is a materially weaker right than the AAT's prior review of plan adequacy. It is a continuity concern that sits alongside, but needs to remain separate to the measurement-validity question.

The questions that need to be on the table

Several questions follow from the gaps above. They are the questions that providers, peaks, and participants should be putting into the co-design and subordinate-rules conversations:

• Does the I-CAN v6 environmental circumstances questionnaire actually separate intrinsic functional reduction from environmental disablement, or does it conflate them?
• How will the design prevent equity failure modes for CALD and First Nations participants, where the score will reflect interpreter mediation and cultural framing as much as function?
• How will the assessment-window problem be resolved for psychosocial and fluctuating-condition participants, whose function varies materially across days?
• How will the design respond to the asymmetry of gaming risk, where coaching from professional advocates favours participants who can afford them and reproduces the inequity the reform claims to address?
• How will the design avoid the perverse incentive for assessors who are paid per assessment, where volume pressure rewards quick completion and discourages engagement with complex or behavioural presentations?
• How will the budget translation engine, the artefact that converts a score into a plan, be made visible and contestable rather than embedded in agency IT?

These are the questions that will determine whether the 2026 design is a different proposition from the 2021 proposition or the same proposition in better clothes.

What this asks of providers now

There are two operational consequences for providers, one strategic posture, and one harder shift.

1. Operationally, providers should expect substantial volume in the reassessment process if the proposition proceeds. The 760,000 figure is the cohort being moved through the new gate. Even at modest rates of plan disruption, that is a substantial volume of plan reviews, support adjustments, and participant transitions to absorb. Workforce planning, document management capability, and transition support will all be in increased demand.

2. Operationally, providers should also expect requests for evidence to support functional capacity assessments. Treating clinician evidence, longitudinal records of how participants function in home and community settings, and documentation of how supports have changed function over time will all become evidentiary material in assessment decisions. Providers who maintain that documentation will be in a materially different position to those who do not.

3. Strategically, I encourage providers to not focus on supporting or opposing the proposition but truly understanding what is actually being proposed. The proposition is currently more rhetorical than designed. Engaging with the rhetoric does not test the design. Engaging with the design at the level of subordinate rules and co-design submissions is where the structural questions can actually be put. The accountability instruments that can shape the design are the JCPAA inquiry, the Commonwealth's own consultation processes, and the parliamentary scrutiny of disallowable instruments. All three are still open.

4. The harder shift is internal, and it is likely going to be uncomfortable for good providers to sit with. Years of task-completion-oriented support design may have suppressed participant capability in ways that the assessment will then record as the participant's baseline. It is unlikely any assessment will be able to tell the difference between intrinsic reduction and a capability that the support model itself has muted. Whichever way the validity argument lands, providers who are still designing supports around task completion rather than capability development have a window, before assessment, to look hard at whether their model is doing the participant any favours.

This is not a comfortable claim and it should be qualified. There is no longitudinal data showing a clear difference in assessment outcomes between support models, because that data does not currently exist. Naming that absence is part of the call. The validation work that should have happened before this gate is built has not been done.

What we are actually asking for

As with all things NDIS, the devil is in the detail. There is no single mechanism that will standardise disability whilst maintaining human rights. Trying to force an overlay without knowing the foundations below it are solid is going to be costly. To get equality, not sameness, the proposition needs to address its own foundations:

• Replacing subjective and fragmented evidence with a standardised dataset, while preserving longitudinal context.
• Resolving differing levels of diagnostic and clinical skill across the assessor workforce, before that workforce becomes the gate.
• Releasing and contesting the budget translation engine, not just the assessment tool.
• Commissioning the validation study before rollout produces decisions that cannot be unwound.

Providers who want to influence the design will need to engage at the technical level, not at the rhetorical one.

> The government has solved the right problem (process) and left the harder problem (measurement science) untouched. That is the part of the design that still needs to be built before functional assessments can act as the Scheme's gatekeeper.

---

Next instalment: the fraud number that is not a fraud number.

Supporting Potential works with NDIS providers to build systems that deliver quality at scale. Our submission to the JCPAA Inquiry into the Administration of the NDIS consolidates prior recommendations into five structural root causes and maps public submissions to the inquiry against them.