Add your promotional text...

Getting the AI Investment Case Right: How the Best Executives Decide?

Paul Wilson

6/17/20266 min read

Every executive I talk to is under pressure to have an AI strategy.

The board wants one, the market expects one, and half the organisation is already experimenting with tools nobody signed off on. That pressure is real, but it's also where a lot of good money goes to die.

McKinsey's State of Organizations 2026 report, published in February, offers a useful reality check. It found that the biggest barriers preventing organisations from scaling AI aren't technical at all. Regulatory and ethical concerns, organisational complexity, legacy infrastructure and a lack of clear strategy or leadership all rank ahead of the technology itself. Seventy two percent of leaders admit their organisation isn't fully ready for what's coming next.

That number tells you something important. The gap isn't between companies that use AI well and companies that don't. It's between companies that qualify their AI investments properly before committing, and companies that fund a pilot because everyone else seems to be funding one.

What the 2026 data actually shows

Deloitte's latest State of AI in the Enterprise report, released this year and titled The Untapped Edge, surveyed more than 3,200 business and IT leaders across 24 countries. The picture it paints is blunt. Adoption is broadening much faster than integration. Around 37% of organisations are using AI with little or no change to the underlying processes it sits inside. Only about 30% are redesigning important processes around it while leaving the core business model untouched. Fewer still, just 12%, are redesigning at scale with a genuinely new operating model behind the investment.

Perhaps the most telling figure in the whole report is this one. Nearly half of respondents say their organisation introduced AI without redesigning the workflows or roles it was meant to improve. You can spend as much as you like on the tool. If nothing around it changes, the investment case was never real to begin with.

Deloitte expects that gap to start closing under board pressure. Right now, only 4% of organisations report board level AI value reporting. By the end of 2026, Deloitte expects that to become standard practice for public companies and large enterprises, as boards stop accepting cost savings alone and start demanding evidence of strategic value.

McKinsey's companion research on AI trust, published earlier this year, adds the governance dimension. Organisations investing twenty five million dollars or more into responsible AI report significantly higher governance maturity, and they are far more likely to see material financial benefit, including EBIT impact above five percent. Responsible AI investment isn't a tax on innovation. It's the precondition for the value showing up at all.

Proof of Value vs Production

The real investment conundrum isn't whether to spend on AI. It's where to place the money that actually matters.

The smartest executives prioritise high value use cases, the ones that show return within six to twelve months, reduce cost directly, lift revenue, or genuinely improve the customer experience at pace. Everything else waits its turn.

The marketplace right now is loud. Early movers are shifting fast from experimentation into full commitment on long term solutions, and the organisations still sitting on the sidelines are falling further behind with every quarter that passes. The gap between AI movers and AI laggards is widening, not narrowing.

Here's the tension underneath all of it. Depending on which source you read, somewhere between 60 and 80 percent of AI experiments and proofs of value never make it into production. That lines up with what Deloitte found this year, where nearly half of organisations introduced AI without redesigning a single workflow around it, and only 12% restructured at real scale. Most pilots simply stall.

The risk of sitting it out

Sitting out isn't the answer either. Without experimentation, you lose the early signal that tells you what's actually working, what isn't, and where the next dollar should go. The organisations getting this right aren't avoiding risk. They're qualifying it properly, moving deliberately from experiment, to proof of concept, to proof of value, to full production, with a clear decision gate at every stage.

How far you push that pipeline still comes down to appetite. Some organisations have the risk tolerance and the capital to make a genuine first mover bet on an unproven solution. Most don't, and shouldn't pretend to.

What makes this harder than a typical capital investment is that AI itself keeps moving. The technology underneath a use case can shift meaningfully in the time it takes to build a business case around it. A single, fixed, full lifecycle investment against a moving target rarely holds up, unless you've deliberately chosen to be the first mover and priced that uncertainty in from the start.

That's why the best executives don't treat AI investment as a one time decision. They stay active in the cycle, continuously reviewing where to stop, where to go, and where to adjust priorities as the technology and the evidence both evolve.

Why good investment governance is the accelerant, not the handbrake

There's a reflex in a lot of boardrooms to treat governance as the thing slowing AI down. Australia's own experience this year says the opposite.

In March 2026 the Digital Transformation Agency released new guidance to help Commonwealth agencies move beyond isolated AI experiments toward meaningful, scaled adoption. It sits alongside an updated national AI policy that makes foundational AI training mandatory for every Australian Public Service employee, with the first requirement taking effect in June and the rest by December this year. The message from the DTA has been consistent. Governance and capability need to advance together, not one after the other.

It's paying off. Australia was ranked second in the world in the OECD's inaugural Digital Government Outlook 2026, a report examining how governments are scaling trustworthy AI adoption and strengthening investment decision making. Its central theme, almost too neatly, was described as the shift from strategy to execution.

That's the pattern I see everywhere it actually works. Governance isn't a brake pedal. It's what lets an executive commit real money to a real deployment without spending the next twelve months managing risk after the fact.

How the best executives actually decide?

Over 25 years running mission critical programs, including the investment and delivery work I lead on Cisco's Country Digital Acceleration portfolio across Australia, the pattern holds. The executives who get value from AI apply the same rigour to it that they'd apply to any major capital investment.

That means testing genuine market and customer pull, not internal enthusiasm. It means qualifying the opportunity against real strategic goals rather than a generic ambition to do something with AI. It means understanding whether a use case can actually scale before the business case gets signed off, not after the money's already spent chasing a proof of concept that was never going to convert.

I use a four way qualification model to pressure test exactly this before an investment case goes anywhere near a board. Strategic fit, market and customer validation, delivery readiness, and governance maturity. It's the same discipline that turned a national AI portfolio worth more than one hundred million dollars into a program that actually delivered, not just one that launched.

Where this leaves you?

If your organisation is one of the many still redesigning nothing around its AI spend, the problem probably isn't the technology. It's that the investment case was never properly qualified before the money moved. Boards are about to start asking harder questions about AI value this year, and the organisations that can answer them will be the ones that did this work early.

Strategy is easy to talk about. Execution, and the discipline behind a properly qualified investment case, is where it counts.

If your next AI initiative is too important to get wrong, let's talk.

Book a confidential briefing session and I'll walk you through how my rapid four way qualification model can pressure test your investment case before it goes to executives and the board.

**Paul Wilson Strategy to Execute www.paul-wilson.net.au