Most companies are buying AI the way people buy gym memberships in January. Big commitment, good intentions, and — six months later — an expensive subscription nobody quite remembers the point of.
I've watched this pattern for 25 years across transformations from €1m to €1B, and the newest tool never changes the oldest mistake: companies ask "how do we use AI?" before they ask "what result do we actually want?"
AI that doesn't change a decision is just cost with better marketing. So if you want AI to create value instead of consuming budget, stop starting with the technology. Start with the decision. Here's the four-step way through it.
1. Name the result you want — not the tool. "We want to use AI" is not a goal. "We want to cut quote turnaround from three days to three hours" is. The tool is a means; the outcome is the mandate. If you can't state the outcome in a sentence a customer would care about, you're not ready to spend.
2. Find the decision that leaks money. Every business has one: a decision made too slowly, too often, or on too little information. Pricing. Credit. Prioritisation. Triage. That leak — not the org chart, not the tech stack — is where AI belongs. Point it there and the value is obvious. Point it anywhere else and you're decorating.
3. Change the call, not the company. Here's the liberating part: you almost never need an "AI team," a reorg, or an eighteen-month platform programme. You need one better decision, made a thousand times. Improve the single call that repeats across your business and the compounding does the rest. Small surface area, enormous leverage.
4. Measure the decision, not the model. Nobody outside your data team cares about model accuracy. They care about faster, cheaper, or better. Pick one of those three, put a number on it before you start, and you'll always know whether the AI earned its keep. No metric, no mandate.
Do you see the shift? Every one of these steps drags the conversation away from "what can this technology do?" and toward "what decision are we trying to win?" That single move is the difference between the companies quietly compounding an edge with AI and the ones quietly writing it off.
This is also, incidentally, why so much AI spend disappears without a trace. It was never attached to a decision, so it could never change an outcome, so it could never show a return. The technology wasn't the problem. The absence of a decision was.
There's a name for doing this systematically — treating transformation as a set of interlocking decisions rather than a shopping list of tools. I call it Total Change Management, and it's the architecture behind every programme we run. But you don't need the framework to start. You need one question, asked honestly, in your next AI meeting:
"Which decision are we actually trying to change?"
If the room goes quiet, you've found the real work.
So let me ask you the same thing: if you could point AI at exactly one decision in your business, which would it be? Drop it in the comments — I read every reply, and the answers are usually more revealing than any strategy deck. 👇