The program closed on schedule. The delivery partner reported the outcomes as achieved. The board accepted the closure paper, the executive sponsor signed off, and the next quarter’s strategic agenda moved on. Twelve to eighteen months later, the metrics that were supposed to have shifted have not. The pattern holds for transformation programs of any kind. It sharpens after an AI transformation, because the technology is easier to blame than the framing that produced the brief. The conversation now is upward — to the board, to the minister, to the CEO — and the question is the one no closure paper anticipated. What actually happened, and what an independent review of the program would actually find.
In Brief
- An independent review initiated after a transformation program rarely names execution failure as the cause. It names the constraint that was misidentified at the start.
- A delivery partner cannot independently review work that implicates the methodology it sold; the structural position rules out the diagnosis.
- Senior buyers in Australia are now initiating a standalone independent review as a governance instrument, separate from any next phase of work.
- Separating the diagnostic question from the investment question, before the next phase is approved, gives the executive strategic clarity their peers do not have.
This is not unusual. It is the typical outcome. The Australian management consulting market is now valued at USD 9.43 billion, growing at 6.07% annually, with the government and public sector segment the largest single category of buyer at 18.24% of total spend (Mordor Intelligence, 2026).
Inside that market, demand for genuinely independent advisory work is rising alongside the broader category. Senior buyers have started separating two questions that used to be answered by the same firm: what was delivered, and what the constraint actually was.
This is not a delivery problem
When a major program does not produce the outcomes it promised, the instinct inside the organisation is to look at execution. Timelines slipped. Scope was compromised. The change management stream was under-resourced. These observations are accurate often enough to feel like an answer, and insufficient often enough that they have become the standard explanation in most post-implementation reviews. BCG’s primary research across more than 800 large organisations finds that only around 30% of transformations meet or exceed their targets and deliver sustainable change — meaning roughly 70% fall short — and the root causes sit upstream of execution: in the framing of the problem, the conviction held inside the leadership team, and the change narrative the program was operating against (Forth et al., 2020).
What surfaces in independent reviews of these programs is consistent. The constraint that limited the outcome was visible before the work began. It was visible in the operating model, in the governance design, or in the accountability structure that surrounded the work. It was not in the delivery team’s capacity to execute against a plan. The delivery team did execute the plan. The plan addressed a problem that was not the actual constraint. The gap between what the program produced and what the executive expected is therefore a diagnostic gap, not a delivery gap. That distinction is the one most closure papers do not make. It is also the one the board is now asking about.
The reason this pattern is so consistent is structural, not accidental. The program was scoped by the same parties who would deliver it. The problem definition and the proposed solution came from the same analytical frame — which made the misidentification almost invisible until outcomes failed to materialise. The executive who approved the brief had no instrument to assess the diagnostic quality of the framing. Only the delivery quality of the execution was visible from that position. This is not a failure of diligence. It is a structural property of how programs tend to be scoped: the people best placed to design the solution are the same people whose frame determines what the problem is.
Why the delivery partner should not review their own work
When executives ask the firm that built the program to review what went wrong, the answer comes back in language that does not implicate the approach. Integrity is not the variable here. Structural position is. A reviewer whose continued engagement depends on the continuation of the relationship is not in a position to identify the constraint that would disrupt it.
The Australian National Audit Office has documented the conditions under which advisory procurement relationships produce accountability gaps. The Auditor-General’s review of the Department of Home Affairs’ procurement of advisory services for the Permissions Capability found that probity, conflict-of-interest, and value-for-money assessments were not adequately managed when multiple advisory firms were engaged with overlapping scopes (Australian National Audit Office, 2023). The same structural logic applies to post-program review. If the firm that designed and delivered the program is also the firm reviewing why it did not work, the answer is constrained before the review begins. The board may receive a thorough document. It will not receive an independent diagnosis.
An independent review asks what the delivery firm cannot
An independent review — initiated by the executive, run by a firm with no stake in the next phase — asks the question the delivery firm is not in a position to ask. It names the constraint that limited the outcome, not the execution failure that is easier to document. The constraint is, in most cases, upstream of the delivery work. It sits in the way the problem was framed at the outset, in the operating model the program was asked to operate inside, or in the governance structure that scoped what the program was permitted to address.
Naming the constraint correctly does not absolve the executive of responsibility for the outcome. It does the opposite. It produces a defensible account of why the program produced what it produced, and what the next investment must address differently. An executive’s program board conversation that begins with a defensible diagnosis differs from one that begins with an explanation. The Minister or CEO asking the accountability question is not asking for the program to be relitigated. They are asking for the diagnosis that explains why the next investment will produce a different result.
AI failure looks technical from outside
The failures inside an AI program are easy to describe in the language of technology: the model hallucinated, the data was not clean, the retrieval layer returned the wrong documents, the prompt engineering was underdeveloped. These descriptions populate the closure paper. They name what stopped working, but they do not explain why the program was scoped to depend on those components in the first place. A closure paper written in that vocabulary closes the discussion inside the program’s own frame, and the question of whether the program was aimed at the right problem does not get asked.
The same firms that scoped the AI transformation also defined what “AI-ready” meant for the organisation and which use cases would carry the investment case. They defined what “delivered” would look like at the end. That is the same structural independence problem that limits any post-program review by the delivery partner. The AI variant is harder to detect because AI success metrics are more definable by the firm that sold the program than the outcome-level metrics the executive originally wanted to move. Model accuracy against a curated test set is one example; adoption within a pilot cohort is another. Both are numbers the delivery firm sets and reports on. A program can meet every one of those internal metrics and still not change the operational condition the executive was investing to change.
McKinsey’s AI Transformation framework is a program-design framework covering the operating dimensions of running an AI program at scale (QuantumBlack, AI by McKinsey, 2026). It describes how to run and scale the program. It does not describe how to diagnose why the outcomes the executive was investing to change did not move. That is a different question, and it does not belong to the firm that ran the program. The executive who asks it independently ends up with an account of the constraint that the delivery firm was not in a position to identify.
The diagnostic question belongs before scoping
The decision to initiate an independent review is not a confession. It is a governance decision, taken by an executive who recognises that the next conversation upward will be stronger if it begins from a defensible diagnosis of the constraint, rather than from a defence of the work that has already happened. The reviewer’s structural independence is what makes the diagnosis usable in front of the board. The diagnosis is what makes the next investment defensible.
Without that separation, the diagnostic question does not disappear. Each program that closes without one carries the unresolved constraint into the next scope. The gap between what the board was told about the previous program and what an independent diagnosis would have shown widens with every cycle that closes without one. A second program that does not move the numbers is not twice as difficult to account for. It is categorically more difficult — because the pattern is now visible to everyone asking.
The Australian advisory market has matured to the point where a standalone diagnostic engagement, distinct from any subsequent program of work, is a recognised governance practice; government and public-sector buyers are the largest segment (Mordor Intelligence, 2026). The firms that designed the original program are not, by virtue of their position, able to provide it. The board is asking the question. The question deserves an answer that has been independently arrived at.
The board question does not get easier with a second program. It gets harder with every one that closes without a diagnosis.
References
- Australian National Audit Office. (2023). Auditor-General Report No. 34 of 2022–23: Procurement of the Permissions Capability. Commonwealth of Australia. https://www.anao.gov.au/sites/default/files/2023-06/Auditor-General_Report_2022-23_34.pdf
- Forth, P., Reichert, T., de Laubier, R., & Chakraborty, S. (2020). Flipping the odds of digital transformation success. Boston Consulting Group. https://www.bcg.com/publications/2020/increasing-odds-of-success-in-digital-transformation
- Mordor Intelligence. (2026). Australia management consulting services market — size, share and analysis. https://www.mordorintelligence.com/industry-reports/australia-management-consulting-services-market
- QuantumBlack, AI by McKinsey. (2026). AI consulting: How we help clients. McKinsey & Company. https://www.mckinsey.com/capabilities/quantumblack/how-we-help-clients
