AI readiness assessment
A five-dimension framework Australian businesses can score in an hour — and a map of the fractional AI leadership shape that fits the result.
· 11 min read
What an AI readiness assessment actually is
An AI readiness assessment answers one question: can this business take an AI initiative from idea to production and keep it running safely? Not whether it has bought the tools. Whether it can finish.
Most assessments circulating in the market are lead magnets. They are free, they take four minutes, and they conclude that the business needs the licence the vendor is selling. That is not an assessment; it is a qualification form. A useful assessment is uncomfortable, names the dimension where a leader changes things first, and is capable of recommending that the business stop starting things.
The stakes are not theoretical. MIT’s Project NANDA research found roughly 95% of enterprise generative AI pilots delivered no measurable P&L return. The failures cluster around organisational readiness — integration, ownership, workflow — far more than around model quality.
The five dimensions
Score each dimension 1 (needs a leader), 2 (leader + owner) or 3 (internal owner). Answer against evidence you could show a board, not against intention.
01
Data foundations
Can we get trustworthy data to a model without a six-month project?
- 1 · Needs a leader
- Critical data lives in spreadsheets and one person's head; no owner, no lineage, no access control.
- 2 · Leader + owner
- Core systems are integrated but data quality is inconsistent and access requires a manual request each time.
- 3 · Internal owner
- Named data owners, documented lineage for the systems that matter, and access governed by role rather than by favour.
02
Use-case clarity
Can we name the business number each initiative is supposed to move?
- 1 · Needs a leader
- A list of AI ideas with no owner, no baseline and no agreed measure of success.
- 2 · Leader + owner
- A prioritised list exists, but the value cases are qualitative and nothing has been killed.
- 3 · Internal owner
- Every live initiative maps to a measurable before-and-after, and the stop-list is as maintained as the roadmap.
03
Governance and risk
If a model produced a harmful or non-compliant output tomorrow, what happens?
- 1 · Needs a leader
- No acceptable-use policy, no record of which AI tools staff are already using, no privacy assessment.
- 2 · Leader + owner
- A policy exists but is not enforced, and risk is assessed per project rather than across the portfolio.
- 3 · Internal owner
- Accountability is documented, higher-risk uses have controls and human oversight, and privacy obligations are assessed before deployment rather than after.
04
Capability and change
Will people actually use what we deploy, and can we run it without the vendor?
- 1 · Needs a leader
- Enthusiasm concentrated in one team; no training, no internal ability to maintain what is built.
- 2 · Leader + owner
- Broad tool access and some training, but adoption is uneven and support depends on an external party.
- 3 · Internal owner
- Role-specific enablement, a named internal owner for each production system, and change managed as part of delivery.
05
Ownership
Who decides what gets funded, what gets killed, and who is measured on the result?
- 1 · Needs a leader
- AI is everybody's side project. Decisions escalate to a committee and stall there.
- 2 · Leader + owner
- A sponsor exists but has no mandate over budget or delivery, and no dedicated time.
- 3 · Internal owner
- One senior person holds the portfolio, has authority to stop work, and is accountable for a business outcome, not a deliverable.
Reading your score
Total the five dimensions for a score out of 15. The total matters less than the distribution: one dimension sitting at 1 is usually where the right fractional leader starts.
01 / 05
Data foundations
Can we get trustworthy data to a model without a six-month project?
Answer against evidence you could show a board, not against intention. Nothing you select leaves your browser.
Prefer a shareable link? The assessment has its own page.
| Score | Reading | What to do next |
|---|---|---|
| 5–9 | Foundation first | A fractional Chief AI Officer should own the portfolio now. The first job is to stop stalled pilots and build one reliable chain from data to decision. |
| 10–14 | Scale with support | A fractional leader paired with an internal owner is the right shape: they own the dimension where execution stalls while the team builds muscle. |
| 15 | Accelerate | The foundations are in place. A fractional advisor can keep governance, risk and roadmap ahead of the pace you scale. |
The Australian context
The governance dimension is where the assessment stops being generic. Australian organisations are now expected to work to the National AI Centre’s Guidance for AI Adoption, published in October 2025, which sets out six essential practices for safe and responsible AI governance. It evolves the earlier Voluntary AI Safety Standard and its 10 guardrails, the first of which is establishing and publishing an accountability process — the ownership dimension, written into national guidance.
Where AI systems touch personal information, the OAIC’s guidance on privacy and commercially available AI products sets the expectation for how Privacy Act obligations apply — including to tools staff adopted without a procurement process. And the Australian Responsible AI Index 2025, produced by Fifth Quadrant for the National AI Centre, tracks how far local organisations still have to travel on exactly these practices.
Adoption itself is no longer the differentiator. The ABS reports that business adoption of AI accelerated through 2024–25, and the National AI Centre’s AI Adoption Tracker follows the same trend. Nearly everyone has started. Readiness is about who can finish.
The pattern in the results
Run this across mid-market businesses and the shape repeats: reasonable scores on tools and use-case enthusiasm, gaps on data access, and the clearest leadership gap on ownership. AI sits with whoever has the most curiosity rather than the most authority, and every decision that requires a trade-off waits for a committee.
That is why readiness gaps are usually leadership gaps. A business turning over A$20M to A$200M feels the same board pressure as an enterprise with none of the payroll headroom for a full-time executive. We set out the cost of that seat in the Chief AI Officer role in Australia and the part-time alternative in fractional AI leadership in Australia.
What to do in the first 90 days
Weeks 1–3: Inventory and stop-list
List every AI initiative running formally or informally, including tools staff adopted without telling anyone. Map each to a business number and kill the ones that map to nothing.
Weeks 3–6: One use case to production
Take the smallest initiative with a real owner and a measurable baseline all the way through — integration, monitoring, rollback, training, and a named person accountable when it breaks.
Weeks 6–10: Guardrails people can follow
Acceptable use, data handling, privacy assessment and human oversight for higher-risk uses, written so staff will actually read them.
Weeks 10–13: Cadence and the ownership decision
A funded pipeline, a monthly executive review, capability gaps named, and an honest call on whether the business needs a permanent hire or another two quarters of senior part-time ownership.
Common questions
- What is an AI readiness assessment?
- A structured review of whether a business can take an AI initiative from idea to production and keep it running safely. It covers data foundations, use-case clarity, governance and risk, capability and change, and — most often the weak point — who owns the portfolio and its outcomes.
- How long does an AI readiness assessment take?
- A self-scored assessment across the five dimensions takes about an hour with the right people in the room. A facilitated assessment with evidence gathering, system access and interviews typically runs two to four weeks.
- Who should run the assessment?
- Someone accountable for the result rather than for selling the next step. Assessments run by a vendor whose product is the recommended remedy tend to conclude that the remedy is the product. An internal executive sponsor or an independent senior operator avoids that conflict.
- Do we need a consultant to assess AI readiness?
- No. The first pass should be internal and honest. External help is worth paying for when the business needs the assessment turned into a funded plan and someone to own delivering it, which is a leadership engagement rather than an advisory one.
- What should we do if we score low?
- A low score is a map, not a verdict. Low scores on data and capability are normal and fixable in sequence. A low score on ownership is the one that stalls everything else, because no other dimension improves without a named senior person deciding what gets funded, killed and carried into production.
- Is AI readiness different in Australia?
- The technical dimensions are universal; the governance dimension is local. Australian organisations are expected to work to the National AI Centre's Guidance for AI Adoption and the Voluntary AI Safety Standard, and to handle personal information in AI systems consistently with OAIC guidance under the Privacy Act.
Where we fit
Chief Orchestration Officer is building Australia’s vetted bench of fractional AI leaders — the people who close the ownership gap without a full-time hire. Practitioners are assessed against a public bar before anything is listed, and the vetting is independent of payment. The founding bench is being vetted now.
Sources
- National AI Centre (Department of Industry, Science and Resources), “Guidance for AI Adoption” — 6 essential AI practices, published 21 October 2025
- Department of Industry, Science and Resources, “Voluntary AI Safety Standard: the 10 guardrails”, published 5 September 2024
- Office of the Australian Information Commissioner, “Guidance on privacy and the use of commercially available AI products”
- Australian Bureau of Statistics, “Business adoption of Artificial Intelligence accelerates in 2024–25”, Characteristics of Australian Business
- National AI Centre, AI Adoption Tracker
- MIT Project NANDA, “The GenAI Divide: State of AI in Business 2025”, July 2025
- Fortune, “MIT report: 95% of generative AI pilots at companies are failing”, 18 August 2025
- Fifth Quadrant for the National AI Centre, “Australian Responsible AI Index 2025”, 26 August 2025
Related reading: Fractional AI leadership in Australia · The Chief AI Officer role in Australia