Don't Get Distracted by AI

In 1970, Pininfarina rolled the Ferrari 512 S Modulo onto the stand at the Geneva Motor Show: a wedge so low you stepped over it, with a glass canopy that slid forward to let you in.

It was never built to be sold. Judged on units shipped, the Modulo is a total failure. Zero.

And yet it won 22 design awards, still anchors the Museo Pininfarina, and shaped the Ferraris that followed. Nobody lost their job over the sales numbers, because sales were never the point. The Modulo was measured on what it taught and what it made possible.

Most companies are about to make the opposite mistake with AI.

The split I saw at the CMO Summit

Last week at the CMO Summit in Amsterdam, a clean split opened up in the room. On one side, the work that has always defined marketing: brand, demand, customer experience. On the other, an agenda borrowed from RevOps and the CRO: pipeline, attribution, and above all AI, AI, AI.

My message to the room was simple. Don't get distracted. Stick to your original KPIs. The fundamentals did not disappear because a new technology arrived.

This is not a "go slow on AI" argument. It is an argument about what you measure, and when.

Is AI spend already running ahead of readiness?

Yes, and the data backs the worry. Gartner's 2026 CMO Spend Survey found CMOs now allocate 15.3% of the marketing budget to AI, while only 30% say they are ready to scale it. Spend has run ahead of readiness. Brand, meanwhile, remains the number one CMO priority for the second year running.

Then there is the money. Uber admitted it burned through its entire 2026 AI budget by April, and the number of teams created purely to manage AI spend has doubled in a year. I wrote in the FD that companies were exhausting their 2026 AI budgets in the first quarter. That is not a spreadsheet-discipline problem. It is structural: you cannot budget for what you do not know yet. I would happily forecast my token spend for the year, until the day a model provider changes the rules and my assumptions evaporate overnight.

So if you cannot forecast it, and you cannot yet prove it, why are we forcing AI into a frame built for things we can forecast and prove?

How should CMOs measure AI? Match the metric to the maturity

Here is the reframe. The Modulo had KPIs. None of them were units sold.

Treat AI the way a car company treats a concept: put it in the laboratory, not on the production line. There the right measures are not revenue and ROI but learning. How fast you can test an idea, how quickly you can kill a bad one, how many assumptions you have proven or disproven, whether people adopt it, and what your organisation can now do that it could not before.

Scott Brinker and Frans Riemersma make the same distinction in their Martech for 2026 work, splitting the stack into the Factory, which protects today's revenue, and the Laboratory, where you experiment. Their data shows more than 90% of organisations now use AI agents, but barely a fifth have them in production. Most of this is still concept-car work. So measure it like concept-car work.

This is the discipline, not the absence of it. A revenue KPI on a prototype is not rigour, it is a category error. It kills the experiment before it can teach you anything, and pushes teams toward safe, unambitious pilots that can show a number this quarter.

Why the value keeps sliding downward: Value Gravity™

This is where Value Gravity™ comes in, because it explains why the distraction is so seductive.

The flashy top layer, the copilots and agents, is where the excitement lives. But it is low-mass: value there commoditises fast, because everyone gets the same models at the same time. The real mass sits lower, in the dense, governed layers: your identity graphs, your consent frameworks, the lifecycle data nobody else has. That is where value accretes and sticks.

A revenue-now KPI is a magnet pointed at that top layer. It pulls attention, budget and talent toward the place where advantage evaporates fastest, and away from where it compounds. "Don't get distracted by AI" really means "don't let the brightest object in the room pull you off the layer where your value lives."

Where does this argument stop?

One line matters. This is an argument about the marketing function and its tools, not about your core business.

If the disruption is at the tool or channel level, you are still playing the same game. Buyers start asking an AI assistant instead of Google, fine. You swap instruments and carry on. Your goals do not change, only your toolkit does.

But if your core product is genuinely being disrupted by AI, that is a different league, and this article is not about it. That company should reorganise around the new reality; telling it to ring-fence 5% and carry on would be malpractice. Know which situation you are in. For most established companies, the real risk is not disruption. It is distraction.

So what should a CMO actually do?

Keep delivering your original mandate. Brand, demand, experience, product. Those KPIs did not expire.

Ring-fence a fixed slice of budget for AI and accept, up front, that you cannot forecast its return. Treat it as a laboratory line, not a revenue line.

Measure that work honestly, on learning rather than earning. A prototype that teaches you something true is a success, even if it never ships.

And hold your nerve when the room fills with people insisting AI should carry a number this quarter. The Modulo never carried a sales number. It is still in the museum. The cars built to chase the quarter are long gone.