How to Build an NDIS Progress Note Reviewer with AI

Progress notes are the single most scrutinised document type in NDIS audits. They're the primary evidence that supports are being delivered, that participants are making progress toward their goals, and that your organisation is meeting the NDIS Practice Standards.

Most providers know this. Most providers also know their progress notes aren't good enough. The same issues appear across the sector: vague language, copy-paste entries, worker-centric descriptions instead of participant-focused narratives, and notes that would tell an auditor nothing about whether the participant actually benefited from the support.

This guide shows you exactly what makes a good progress note versus a bad one, provides the criteria you need for systematic quality checking, and walks you through building an AI reviewer that scores progress notes before they're filed — catching quality issues when they can still be fixed.

What Makes a Good NDIS Progress Note

The difference between a good and bad progress note is specific and measurable. Here's a direct comparison.

Bad Progress Note

"Assisted Sarah with daily living activities. She had a good day. No issues to report."

This note tells an auditor nothing. What activities? What does "good day" mean? What was Sarah's response? How does this connect to her plan goals? An auditor reading this would flag it immediately.

Good Progress Note

"Supported Sarah with her morning routine goal (Plan Goal 2: Increase independence in personal care). Sarah completed 3 of 4 steps in her morning routine independently — choosing clothes, brushing teeth, and eating breakfast. She requested verbal prompting for hair brushing, which is consistent with last week. Sarah expressed pride in choosing her own outfit ('I picked this one myself'). Duration: 1.5 hours, 7:00-8:30am."

This note demonstrates: person-centred language (Sarah's perspective and choices), measurable outcomes (3 of 4 steps), plan alignment (Goal 2), participant voice (direct quote), and completeness (duration, time).

The Quality Criteria

Common Progress Note Failures

These are the patterns that auditors — and quality managers — see most frequently.

1. Copy-Paste Notes

The same note, word for word, appearing across multiple days or even multiple participants. This is an immediate red flag for auditors because it suggests either the support wasn't individualised or the notes are fabricated.

Red flag example:

Monday: "Supported client with community access. Client engaged well. No concerns." Tuesday: "Supported client with community access. Client engaged well. No concerns." Wednesday: "Supported client with community access. Client engaged well. No concerns."

Fix: Every note must describe what actually happened that day — specific activities, specific responses, specific outcomes. If Tuesday's session genuinely was identical to Monday's, describe what was the same and why.

2. Worker-Centric Language

Notes that describe what the worker did rather than what the participant experienced.

Worker-centric: "I took John to the shops and helped him buy groceries. I prompted him to use his list."

Person-centred: "John visited Woolworths to practise his independent shopping goal. He used his picture shopping list to find 4 of 5 items independently. John asked for support locating the bread aisle — the store had recently rearranged. He paid at the self-checkout with verbal prompting for the card tap."

The difference: the second version centres John's experience, choices, and agency. The first version centres the worker's actions.

3. Vague Outcome Language

Words and phrases that sound like progress notes but communicate nothing specific.

4. Missing Plan Goal Connection

Notes that describe activities without connecting them to the participant's NDIS plan goals. An auditor needs to see that supports are goal-directed — not just activities for the sake of filling time.

Without goal connection: "Went to the park and practised throwing a ball."

With goal connection: "Attended Riverside Park as part of Goal 3 (Build social connections in the community). Practised ball skills with two regular park group members. Initiated conversation with one peer — the first time Marcus has initiated without prompting. This connects to his goal of building independent social connections."

5. Insufficient Detail for High-Risk Situations

Incidents, behavioural events, and restrictive practices require detailed documentation. Brief notes in these situations are a serious compliance risk.

Insufficient: "Client became upset. Situation resolved."

Required level of detail: "At 2:15pm, James became distressed during the transition from lunch to afternoon activities (consistent with his behaviour support plan trigger: unexpected schedule changes — the art supplies hadn't arrived). James raised his voice and moved away from the group. Staff followed the de-escalation protocol in his BSP: gave James space, offered his calm-down box, and used the visual schedule to show what would happen next. James used his stress ball for approximately 5 minutes, then re-engaged with a modified activity (drawing instead of painting). No restrictive practice was used. James's behaviour support practitioner was notified by email at 2:45pm. Duration of event: approximately 15 minutes."

Building Your Progress Note Review Criteria

Before building a reviewer, you need clear criteria. These criteria should be documented, shared with all staff, and used consistently.

Criteria Set for Standard Progress Notes

1. Person-Centred Language (Weight: 3 — Critical)

The note must:

• Use the participant's name (not "client" or "consumer" throughout)
• Describe the participant's experience, choices, and preferences
• Include the participant's voice where possible (direct quotes, expressed preferences)
• Focus on what the participant did, not what the worker did
• Respect dignity — no patronising language ("was well-behaved", "complied with instructions")

2. Measurable Outcomes (Weight: 3 — Critical)

The note must:

• Include specific, observable details (numbers, durations, frequencies)
• Describe what the participant achieved, attempted, or declined
• Compare to previous sessions where relevant ("improvement from last week", "consistent with recent sessions")
• Avoid subjective judgements without supporting evidence
• Enable someone reading the note to understand what actually happened

3. Plan Goal Alignment (Weight: 2 — Important)

The note must:

• Reference the specific plan goal the support relates to
• Explain how the activity connects to goal progress
• Note any changes in goal progress (forward, stable, or regression)
• Flag when support may no longer align with current goals

4. Completeness (Weight: 2 — Important)

The note must include:

• Date of service delivery
• Start and end time (or duration)
• Worker name
• Support type/category
• Location (where relevant)
• Participant's response to the support
• Any follow-up actions needed

5. Professional Standards (Weight: 1 — Standard)

The note must:

• Be written within 24 hours of service delivery
• Use current NDIS terminology
• Be free of spelling and grammar errors that affect clarity
• Not contain personal opinions about the participant
• Not include information about other participants

Step-by-Step: Building the AI Reviewer

Here's how to create a progress note reviewer in TeamBench that scores notes against the criteria above.

Step 1: Create the Reviewer

Set up a new AI reviewer with these settings:

Name: NDIS Progress Note Quality Checker

Description: Reviews NDIS participant progress notes for person-centred language, measurable outcomes, plan goal alignment, completeness, and professional standards. Designed for support workers, team leaders, and quality managers.

Step 2: Write the System Prompt

The system prompt tells the AI exactly what to check and how to provide feedback.

You are an NDIS progress note quality reviewer for an Australian disability services provider. Review each progress note against the NDIS Practice Standards for provision of supports.

Check for:

1. PERSON-CENTRED LANGUAGE: Is the note written from the participant's perspective? Does it use their name? Does it describe their experience, choices, and voice? Flag worker-centric language ("I assisted", "I took them") and suggest participant-centred alternatives.
2. MEASURABLE OUTCOMES: Does the note include specific, observable details? Numbers, durations, frequencies? Can someone reading this note understand what actually happened? Flag vague language ("good day", "engaged well", "no issues") and suggest specific alternatives.
3. PLAN GOAL ALIGNMENT: Does the note reference a specific plan goal? Does it explain how the support connects to goal progress? Flag notes that describe activities without connecting to goals.
4. COMPLETENESS: Are all required elements present — date, time/duration, worker name, support type, participant response, follow-up actions?
5. PROFESSIONAL STANDARDS: Is the note timely, uses current NDIS terminology, free of errors, and objective?

For each issue found, quote the specific text, explain why it's a problem, and provide a concrete rewrite suggestion. Be constructive — the goal is to help support workers write better notes, not to criticise them.

IMPORTANT: Do NOT flag copy-paste detection for a single note — that requires comparing across multiple notes. Focus on the quality of this individual note.

Step 3: Set Evaluation Criteria

Add five evaluation criteria with the weights from the criteria set above:

1. Person-Centred Language (weight: 3) — Written from participant's perspective, uses their name, includes their voice and choices
2. Measurable Outcomes (weight: 3) — Specific, observable details demonstrating progress or participation
3. Plan Goal Alignment (weight: 2) — Explicitly connects support to participant's NDIS plan goals
4. Completeness (weight: 2) — All required elements present (date, time, duration, worker, support type, response)
5. Professional Standards (weight: 1) — Timely, correct terminology, objective, error-free

Step 4: Set the Quality Gate

Set a minimum score of 70/100. Notes below this threshold are flagged for revision before filing.

Why 70? It's achievable for competent staff while still catching genuinely poor notes. You can raise this threshold over time as your team's note quality improves.

Step 5: Add a Knowledge Base

Upload these documents to a Knowledge Base and attach it to the reviewer:

• Your organisation's progress note policy and procedure
• NDIS Practice Standards — Provision of Supports module
• Your organisation's person-centred language guide (if you have one)
• Sample progress notes graded as "excellent", "acceptable", and "needs improvement"
• Your participant plan goal templates or goal framework

This gives the reviewer context about your specific requirements — not just generic NDIS standards.

Step 6: Test with Real Notes

Before rolling out, test the reviewer with 10-15 real progress notes:

• 3-5 notes your quality team considers "good"
• 3-5 notes your quality team considers "needs improvement"
• 3-5 notes that are clearly "poor"

Check that the reviewer's scores align with your quality team's judgement. If good notes score below 70 or poor notes score above 70, adjust the system prompt or quality gate threshold.

The Review Workflow

Once your reviewer is configured and tested, implement this workflow:

For Support Workers

1. Write the progress note as usual — within 24 hours of service delivery
2. Submit for AI review — paste or upload the note
3. Review the scored feedback — see your overall score and per-criteria breakdown
4. Improve if needed — use the specific feedback to revise the note. Use "Improve with AI" for a one-click rewrite based on the feedback, then review the suggested changes
5. Re-submit until the quality gate passes (score ≥ 70)
6. File the note once it passes the quality gate

For Team Leaders and Quality Managers

1. Review passed notes — spot-check a sample of notes that passed the quality gate (you shouldn't need to review every note)
2. Review flagged notes — any notes that don't pass the quality gate after two attempts should be escalated
3. Monitor score trends — track average scores by worker and by team over time
4. Identify training needs — workers who consistently score low on specific criteria need targeted coaching
5. Report to management — quarterly quality reports showing score trends, pass rates, and improvement areas

Expected Outcomes

Based on the patterns we see across content review workflows:

The improvement comes from two sources: workers learn from the specific feedback (the most important factor), and the quality gate prevents poor notes from being filed without revision.

What AI Review Can and Cannot Do

AI Review Is Good For

• Consistency checking — every note is reviewed against the same criteria, every time
• Specific feedback — pinpoints exactly which phrases are vague or worker-centric, with suggestions
• Scale — reviews 50 notes per day with the same rigour as the first one
• Training — the feedback loop teaches writers faster than periodic training sessions
• Documentation — score history provides evidence of your quality assurance process

AI Review Cannot

• Verify that support was actually delivered — a well-written note about a session that didn't happen is still fraud
• Assess clinical appropriateness — whether the support approach was correct for this participant
• Detect cross-note patterns — copy-paste across days requires comparing multiple notes (a separate process)
• Replace clinical supervision — complex situations need human oversight, not just documentation review
• Guarantee audit outcomes — good notes are necessary but not sufficient for passing an audit

Be honest about these limitations with your team. AI progress note review is a quality assurance tool for documentation. It's not a compliance guarantee.

Training Staff to Write Better Notes

The AI reviewer helps, but training is still essential.

Run a Progress Note Workshop

1. Show before/after examples — use real (de-identified) notes from your organisation
2. Score notes together — have the team score notes against the criteria manually, then compare with the AI reviewer's scores
3. Practise rewrites — give staff a "poor" note and 5 minutes to rewrite it. Compare results
4. Discuss common patterns — which criteria does your team struggle with most?
5. Set expectations — the quality gate score, the review workflow, and what happens when notes don't pass

Create a Quick Reference Card

Give every support worker a laminated card or phone-accessible guide:

Every progress note must include:

• ✅ Participant's name (not just "client")
• ✅ What they did, chose, or experienced
• ✅ Specific details (numbers, durations, what happened)
• ✅ How it connects to their plan goals
• ✅ Their response or feedback (in their words if possible)
• ✅ Date, time, duration, your name

Every progress note must avoid:

• ❌ "Had a good day" (what does "good" mean?)
• ❌ "No issues" (describe what went well instead)
• ❌ "I assisted with..." (focus on the participant, not you)
• ❌ "As per usual" (describe what actually happened)
• ❌ Copy-pasting from yesterday

Frequently Asked Questions

How long should an NDIS progress note be?

There's no prescribed length, but a quality note typically takes 100-200 words. The key is specificity, not length. A 50-word note with measurable outcomes and plan alignment is better than a 300-word note full of vague language. Aim for enough detail that someone who wasn't present can understand what happened and how the participant responded.

Can support workers use the AI reviewer on their phones?

TeamBench works in any web browser, including mobile. Support workers can write notes on their phone and submit them for review immediately after a session. The scored feedback helps them improve the note before filing it.

What if a worker's notes consistently fail the quality gate?

This is a training opportunity, not a disciplinary issue. Review their scores to identify which criteria they're failing on — it's usually one or two specific areas. Provide targeted coaching with examples. Pair them with a strong note-writer for a week. If scores don't improve after coaching, escalate to their supervisor.

How does this work with our existing client management system?

The AI reviewer is a quality check that happens before notes are entered into your CMS. Workers write the note, review it with AI, improve if needed, then copy the final version into your system. It adds 2-3 minutes per note for the review step, but saves significant time on quality manager reviews downstream.

Should we review every progress note or just a sample?

AI review can check every note — that's the advantage of automation. We recommend submitting all notes through the reviewer, with the quality gate as the first-pass filter. Quality managers then spot-check a random sample of passed notes (10-15% is typical) rather than reviewing every note manually.

What about notes for incidents or restrictive practices?

These require more detailed documentation and should have additional review criteria. Consider creating a separate reviewer for incident and restrictive practice notes with higher-weighted criteria for detail, de-escalation documentation, BSP alignment, and notification compliance. The quality gate threshold should be higher (e.g., 80/100) given the regulatory risk.

How do we measure whether this is actually improving note quality?

Track three metrics: average first-submission score (should trend upward over months), first-pass rate (percentage of notes that pass the quality gate on first submission — should increase), and quality manager review time per note (should decrease). Report these monthly to management.

Is the AI reviewing participant information? What about privacy?

Any AI-assisted review should comply with your privacy obligations under the Privacy Act 1988 and your NDIS Practice Standards obligations regarding participant information security. Review your organisation's privacy policy and seek advice on appropriate use of AI tools with participant information.

Key Takeaways

• Progress notes are the most scrutinised documents in NDIS audits — they're the primary evidence of service delivery and participant outcomes.
• Five criteria define note quality: person-centred language, measurable outcomes, plan goal alignment, completeness, and professional standards.
• The most common failures are vague language, copy-paste notes, worker-centric descriptions, missing goal connections, and insufficient detail for high-risk situations.
• An AI reviewer can score every note consistently against your criteria, providing specific feedback that helps workers improve.
• Set a quality gate at 70/100 to start — notes below this threshold get revised before filing. Raise the threshold as quality improves.
• Test with real notes first — verify the reviewer's scores align with your quality team's judgement before rolling out.
• Training plus AI review is the winning combination. The reviewer provides ongoing feedback; workshops and coaching build the underlying skills.
• AI review is a quality assurance tool, not a compliance guarantee. It checks documentation quality — it can't verify that support was actually delivered.

---

This article provides general guidance on NDIS progress note documentation quality and is not compliance advice. Always consult the NDIS Quality and Safeguards Commission for current Practice Standards requirements and seek professional guidance for your specific situation.