eLearning design for NDIS support workers: what actually changes practice

Disability support is a different job and it needs different training design. A worker in a participant's home has no supervisor at the desk beside them, no policy folder to flick through, and often no phone signal. Whatever they are going to do, they are going to do it in the next ninety seconds, alone.

That is the test any course in this sector has to pass. Not whether it was completed. Whether it changed what happens at the point of decision.

What follows is what that means in practice. It applies whether a provider builds its own courses, buys them, or asks a training partner to make them.

1. Start with the decision, not the content

Most course design starts with a document: a policy, a practice standard, a set of learning outcomes lifted from a framework. The content gets organised, a quiz is put at the end, and the result is accurate and forgettable.

The better starting question is not "what does this policy say". It is "what is the moment where a worker has to get this right, and what makes that moment hard?"

Take dignity of risk. The content version explains the principle, contrasts it with duty of care, and asks three questions about the definitions. The decision version puts a worker in a kitchen with someone who wants to make their own lunch, a knife, and a support plan that says supervision is required, and asks what they do now.

The second version teaches the same principle. It also rehearses the exact moment the principle gets tested, which is the part the first version leaves entirely to chance.

In practice:

• Write the hardest realistic moment first. Build the teaching around it afterwards.
• If a scenario has an obvious answer, it is not the moment. Find the version where two reasonable workers would disagree.
• Ask an experienced team leader what goes wrong most often. That list is a course outline.

2. Interactivity has to earn its place

Drag and drop, flip cards and hotspots are often decoration. They add clicks, build time and cost, and change nothing about what a worker can do afterwards.

Interactivity is worth including when it forces a decision the worker would otherwise avoid making. It is not worth including when it is a different way to reveal a paragraph.

A useful test: if the interaction can be completed correctly without understanding anything, it is a click, not a decision.

In practice:

• Every interaction should have a wrong answer a competent person might genuinely choose.
• Cut anything that only reveals information. A paragraph does that faster.
• Branching is worth the build cost. Clicking to reveal usually is not.

3. Make the wrong answers plausible

This is the single most skipped step in compliance eLearning, and it is the one that decides whether a course teaches anything.

A question with one sensible option and three obviously wrong ones tests reading comprehension. The worker scores 100%, the dashboard turns green, and nobody has learned that the situation was difficult, because the course never admitted that it was.

Distractors should be the things people actually do. The shortcut taken at the end of a double shift. The well-meant intervention that takes away a choice. The escalation that should have happened two hours earlier and did not.

In practice:

• Source distractors from real incident reports and real supervision conversations, de-identified.
• If nobody picks a wrong answer, the question is not working. Look at the data and rewrite it.
• Never write a joke option. It tells the learner the course is not serious.

4. Feedback should teach the reasoning, not confirm the answer

"Correct. Well done." teaches nothing. "Incorrect. The right answer was B." teaches slightly less than nothing, because it trains the learner to look for the answer rather than think about the problem.

Most of the learning in a scenario happens after the choice is made. That is the moment the learner is most receptive, because they have committed to a position and want to know whether it held.

Good feedback does three things. It says what the choice would have led to. It explains the reasoning that makes the better choice better. And it does that for the right answers too, because a worker who guessed correctly has learned nothing unless they are told why.

In practice:

• Write feedback for every option, not just the wrong ones.
• Explain the consequence, not just the rule. "This is what happens next" beats "this breaches clause 4.2".
• Keep the tone of an experienced colleague at handover, not an auditor. Shame ends learning.

5. Build it around the person supported, not just the role

Role-based training is the sector default. Support worker gets the support worker pathway. It is simple to administer and it is not enough.

Two participants with the same funded supports can need completely different things from a worker. One is autistic and non-speaking with a communication system the worker has to learn. One has epilepsy and a seizure management plan. One has a behaviour support plan with a restrictive practice in it. A generic support worker pathway prepares someone for none of them specifically.

Training that follows the person, rather than only the job title, is the difference between a worker who has done their induction and a worker who is ready for Monday.

In practice:

• Build a short pathway per participant, on top of the role pathway, not instead of it.
• Include the condition-specific and support-need-specific content the plan implies.
• Revisit it when the plan changes. A pathway set once is a pathway going stale.

6. Length: the shift is the real constraint

A course competes with a worker's break, their commute, and the end of a long day. Design for that or the course gets clicked through, which is worse than not running it, because a course that gets clicked through still shows up green on the dashboard.

Under 30 minutes is a realistic ceiling. Mobile-first is not a nice-to-have, because most frontline workers in this sector do not sit at a computer. A course that assumes a desktop assumes a job these workers do not have.

In practice:

• Design on a phone screen first. If it only works on a laptop, it will not be finished properly.
• One idea per screen. Two ideas is a screen that gets skimmed.
• No app store, no install, no separate login to remember. Every step of friction costs completions.

7. Reading level is a design constraint, not a courtesy

A large part of this workforce speaks English as an additional language. Many more are reading tired. Complex sentences do not make training more rigorous, they make it less likely to be understood, and an unclear question tests reading rather than judgement.

CORA holds every course to a readability ceiling and checks it before release, alongside a limit on sentence length and a rule that sentences say the thing directly rather than approaching it. Not because the content is simple, but because the decisions inside it are not, and the language should not be what makes them hard.

In practice:

• Run a readability check on every course before release and treat it as a pass or fail gate.
• Say the condition by name. Euphemisms like "becoming unwell" leave a worker guessing.
• Write literally. If a sentence takes a run-up before it says what it means, cut the run-up.

8. Assessment that measures judgement, not clicks

A completion record answers one question: did this person reach the end. It cannot tell a provider whether the training changed anything, which is the question an auditor, a board and a family will all eventually ask.

The alternative is not a harder quiz. It is capturing the decisions the worker made inside the course, and reporting on those. When each decision is scored and mapped to the relevant NDIS Practice Standard, a provider can see where judgement is strong and where it is thin, by worker and by team, before an incident tells them. CORA scores this against four capability pillars: Capability, what a worker sees and understands. Operational Consistency, what they do the same way every time. Readiness, what they do when it does not fit the pattern. And Assurance, what the provider can be confident about overall.

One boundary matters here. No online course determines whether a worker is competent. That is a judgement a suitably qualified assessor makes after watching real practice. Where a support genuinely requires hands-on assessment, an online course is the preparation and not the proof. What the course can do is make that judgement informed, and give the assessor something defensible to work from.

In practice:

• Report on decisions, not just completions.
• Map to the Practice Standards so the evidence lines up with what the service is assessed against.
• Keep observed assessment for the things that need hands, and say so plainly.

What the evidence says

This is not just a design preference. The comparison has been made properly, in a sector with more research money than this one.

In 2011, a team at Northwestern published a meta-analysis in Academic Medicine asking whether simulation-based education with deliberate practice beats traditional clinical teaching. They screened 3,742 articles from 1990 to 2010 and found 14 that made a direct comparison. Deliberate practice won, with an effect size of 0.71, which is large by the usual conventions.

A related finding is less comfortable. A 2005 systematic review in the Annals of Internal Medicine asked whether doctors improve with time in practice, and concluded that more experienced physicians may paradoxically be at risk of providing lower quality care.

Both are health professions rather than disability support, and nobody has run these studies on support workers. But the mechanism does not belong to medicine. Experience alone does not refine itself. What refines it is being put in a decision and then shown what the reasoning missed.

That matters more in this sector, not less. Most support workers enter the job with no mandated qualification and no structured refinement afterwards. The NDIS Workforce Capability Framework describes what a worker must be able to do, not what they must hold, and that is the right call. Lived experience is the raw material this sector runs on. The question was never whether to value it. The question is what refines it.

Design rules are not an internal detail

Every CORA course is built to the same rules. A course does not ship if it fails them.

That sounds like a process note. It is not. Those rules decide whether what a worker learned reaches the person they support, or stops at a certificate in a compliance folder.

There is a second effect, and providers underrate it.

A worker who finishes a course able to do something new has been invested in. A worker who clicks through one has been processed. They know the difference.

Turnover here is expensive. Recruitment, screening, induction, buddy shifts, then months before a new worker knows a participant well enough to notice when something is off. None of it shows up as a line item, so it rarely gets counted.

Training that treats a worker as worth developing is one of the cheapest things a provider can offer. Most workplaces in this sector do not. That is the opportunity.

The short version

Start from the hardest realistic moment, not from the policy document.

Cut interactivity that does not force a decision.

Write wrong answers a competent person might actually choose.

Explain the reasoning after the choice, for right answers as well as wrong ones.

Build around the participant, not only the role.

Under 30 minutes, mobile-first, one idea per screen.

Treat readability as a release gate.

Report on decisions and map them to the Practice Standards.

Keep observed assessment for what genuinely needs it, and say so.

Completion is the floor. Capability is the point.

CORA Workforce builds NDIS-specific scenario-based courses and holds the workforce record around them. 90+ courses across six streams, each designed to run in under 30 minutes on a phone.