Supporting Potential — Systems Model
An NDIS provider needs eight systems running continuously, each with its own owner and its own cadence. Providers rarely come unstuck inside one of these systems. They come unstuck at the points where two systems have to interlock precisely, and nobody designed the joint.
Where it becomes unstuck
Most complaints are just early signal — minor ones feed Continuous Improvement, serious ones escalate to Incident Management. What survives outside both is the obligation to the complainant: acknowledged, kept informed, told the outcome. Skip that thin layer and people stop telling you what's wrong.
Not a system with its own daily transactions — a governance lens over revenue (claims) against cost (rostered hours, travel, cancellations). Treat it as a separate finance project and you lose the read on which programs are actually viable.
The real-time join between who's capable and available, and what a participant's plan actually requires. Treated as HR admin instead of a designed interface, and you get unfilled shifts, skill mismatches and unfamiliar workers — the leading driver of both quality incidents and cost blowouts.
Safe systems of work and injury/return-to-work sit inside the employee journey; environmental and behavioural hazards sit inside the client journey. Without a mandatory roll-up into Risk and Governance, hazard patterns across the whole workforce stay invisible until something serious happens.
Tested against Achieve Australia
Every feedback loop mapped in this engagement — incident management, funding/billing/claims, the CI register, client lifecycle — independently shows the same shape: the system on either side of a joint works, and nobody was ever assigned the joint itself. Checked against all four fault lines above, the pattern holds four for four, though not at the same severity each time.
Achieve's own Client Entry Procedure lists a Complaints and Feedback Register as one of three "Applicable Registers" governing entry decisions — alongside Continuous Improvement and the Clients Who Live Alone Register — implying complaint signal is meant to feed both CI and Incident Management.
No process document in the evidence base shows how or when that connection actually happens. The register exists; the joint doesn't.
Source: Client Entry Procedure §2, via achieve-client-lifecycle-loop.md; achieve-data-criticality-register.mdReconcile Payments — the step that should close the funding loop — carries no named owner in Achieve's own 2024 process map, labelled "Role TBC." Asked directly in July 2026 whether that had since been resolved, neither the CFO nor the GM Finance & Corporate could name a person.
"Is there a person in the business that actually owns reconciliation?... in the original funding optimization outputs, it said role TBD." The answer described the activities involved. It did not name an owner.
Source: Wally Phillips/Frankie Poon interview, 10 Jul 2026, 19:53–20:21; achieve-funding-billing-claims-loop.mdPay inquiries and rostering duplications are still occurring after the VisiCase uplift completed — a live, unresolved defect, not a hypothetical risk. Asked who owns the fix:
"I don't think you can pinpoint that to the one person. I think it's more the process."
Source: Jay Balana interview, 9 Jul 2026, 6:53; corroborated by the unnamed turnover/attrition rollup step in achieve-board-kpi-reporting-loop.mdLTIFR is the board's most consistently "Behind" workforce KPI (19.3% vs a 12% target, worsening). But no document in the evidence base names the system where a workplace injury is first recorded, or the logic that classifies it as lost-time, before it reaches the WHS dashboard.
This is a level worse than "unowned." The individual event isn't confirmed to be captured anywhere before it becomes a board number.
Source: achieve-board-kpi-reporting-loop.md, LTIFR pipeline findingFour fault lines, four different severities — from a register that exists but isn't wired in, through to a KPI whose underlying event isn't confirmed to be captured at all. None of the four have a continuous owner today. That's not a criticism of any one function: each side of every joint is someone's job and, where evidenced, is being done. The joint itself was never anyone's.