The paradox nobody wants to name
Most businesses have already adopted AI. Almost none of them can point to the productivity gain it was supposed to deliver.
Globally, organisational AI adoption has reached roughly 88%, according to McKinsey’s 2025 State of AI report, while preliminary MIT NANDA research found that 95% of organisations studied reported no measurable P&L return from their GenAI initiatives. Only 1% of business leaders describe their companies as genuinely “mature”, meaning AI is fully embedded in workflows and driving business outcomes rather than sitting alongside them.
In Australia, the picture is consistent. The Productivity Commission estimates AI could add about 0.4 percentage points a year to labour productivity over the next decade. Commonwealth Bank economists estimate a 0.8–1.0 percentage-point annual uplift could be credible for advanced economies as adoption deepens. The Commission’s own framing is instructive: adoption matters, but it is institutional change and the rewriting of existing business practice that moves the number.
This is not a story about whether AI works. The evidence at an individual task level is increasingly clear: AI can make people faster and more productive. The challenge is converting that performance into a measurable business result.
That is the gap this article addresses and it is the reason our approach starts with the business problem, not the technology.
Why “more tools” isn’t the answer
The instinct, when productivity doesn’t move, is to add capability: a new platform, a better model, another integration. The data suggests this instinct is misdirected for most businesses.
Two findings make the case plainly. In Australia, 93% of respondents to the Governance Institute’s 2025 survey said they could not effectively measure AI return on investment. And 42% of companies abandoned most of their AI initiatives before production, as teams moved beyond simple use cases into anything resembling a core workflow.
Neither of these is a technology failure. They are organisational readiness failures and they happen well before a business is in a position to judge whether a tool works.
This is precisely why a business-problem-first approach outperforms a technology-first one, and why we assess readiness across five dimensions before recommending any tooling decision.
The 5Cs: a readiness framework for AI-driven productivity
1. Culture: is change genuinely embedded, or just endorsed?
Adoption figures can be misleading, because usage and trust are not the same thing. In Australia, 57% of employees report relying on AI output without evaluating its accuracy. More tellingly, only 30% of Australians believe AI’s benefits outweigh its risks, despite adoption rates that are among the highest in the world.
This is a culture signal, not a tools signal. A business can hand every employee access to AI and still have no genuine cultural embedding of innovation or change. People will use the tools quietly, inconsistently and without accountability because the organisation hasn’t established what “good” looks like.
The diagnostic question: Is experimentation with AI actively encouraged and structured, or is it happening informally, without oversight, because leadership hasn’t defined how it should be used?
2. Commercial value: is AI improving the business?
You cannot manage what you don’t measure. Yet many businesses using AI still don’t have a clear way of measuring whether it’s actually creating value. As noted above, 93% of respondents to the Governance Institute’s 2025 survey said they could not effectively measure AI return on investment. Where measurement is informal, “did we hit the objective” can become the norm—a subjective, retrospective check rather than a quantified return.
For business leaders, this matters because there is less room for investment that doesn’t deliver a return.
The diagnostic question: If asked today what your business spends on AI tools and what return that spend generates, could you answer with a number or an impression?
3. Capability: can the team actually drive performance from the technology?
Access to a tool is not the same as capability to use it well. AI usage is high, yet many organisations still lack the policies, training and quality standards needed to turn that usage into consistent performance.
This gap compounds by persona. A team member using AI for email drafting needs different capability support than one managing autonomous agents in a production workflow, but most businesses apply a single, generic “AI training” response regardless of where someone sits on that spectrum.
The diagnostic question: Do your teams know not just how to use AI, but how to know when it’s working and is that capability being built deliberately, or assumed?
4. Capacity: are teams set up to sustain this, or is it one more thing on top?
Culture determines whether learning is valued. Capacity determines whether there’s actually room for it. This is the dimension most businesses skip entirely, and the abandonment data shows the cost of skipping it: initiatives that looked promising in pilot stall out not because the technology failed, but because no one had the time, structure, or support to take it from experiment to habit.
The diagnostic question: If you asked your team to adopt a new AI-driven process tomorrow, would they have the bandwidth to actually learn it or would it compete with everything else already on their plate?
5. Commitment: is leadership driving this top-down, or funding it and hoping?
This is the dimension that determines whether the other four ever get addressed. PwC’s CEO Survey found that nine in ten Australian CEOs see AI adoption as central to their business strategy over the next three to five years. Yet strategic intent doesn’t automatically translate into operational change.
For business leaders, this is the uncomfortable finding: it is not enough to declare AI a strategic priority. Commitment shows up in whether management teams are actively empowered with clear mandates, resourcing, and accountability to drive the change, not just told it matters.
The diagnostic question: Is your leadership team declaring AI a priority, or actively resourcing and empowering the change? The difference is whether anyone would notice if it stopped.
Where does this leave your business?
You don’t necessarily need another AI tool.
You need to understand where AI is already being used, where it’s creating value, and what’s getting in the way.
That means looking at:
- where AI is already being used across the business
- whether it’s actually improving productivity or reducing cost
- whether your data and processes are ready to support it
- whether your people have the capability and capacity to use it properly
- and where AI can make the biggest difference next
The goal isn’t to adopt more AI. It’s to make the AI you’re already using or considering work harder for the business.
The takeaway
AI productivity isn’t about having the most tools or the most advanced technology. It’s about whether your business is set up to turn those tools into real results.
Culture, commercial value, capability, capacity and commitment all matter. If one of those is missing, AI can quickly become another cost, another project, or another tool sitting alongside the way your business already works.
Business problem first. Technology follows.