A faster horse
AI can make marketing faster. But if the operating model does not change, the number will not move.
AI can make marketing faster. But if the operating model does not change, the number will not move.
Henry Ford is supposed to have said that if he had asked people what they wanted, they would have asked for a faster horse. Whether he actually said it doesn’t really matter. The point holds. Faced with something genuinely new, most people imagine a faster version of what they already have.
That’s where most AI use cases in marketing are sitting right now.
A faster horse.
The team doubles its content output. The monthly report gets drafted in an afternoon instead of a week. The nurture sequence writes itself. Campaigns ship quicker. The board deck has an AI slide. Everyone can feel the speed, and speed feels like transformation.
But it is still the same machine, running faster.
That feels good for a period of time. Then the number comes in and nothing meaningful has moved. Pipeline quality has not improved. Sales still does not trust the handoff. Conversion rates are static. The same objections keep appearing in late-stage deals. CAC has not changed. The board still cannot see where the commercial system is leaking.
The function looks more productive. And it probably is.
But productive at what?
This is the trap inside most AI adoption in marketing. The work gets faster before anyone asks whether it is still the right work.
The clearest version is the AI experimentation pod. Hand AI to RevOps or marketing ops as a side project. Let them run pilots. Ask them to report back at the QBR.
It looks responsible. It rarely changes the business.
The pod optimises for something demonstrable: a chatbot, a content generator, a workflow automation, another dashboard. Something visible enough for the board deck, but disconnected from the value-creation plan. It proves the company is doing something with AI. It does not prove AI is improving pipeline, conversion, revenue productivity or EBITDA.
That is the move companies make when they want to say AI has a mandate without giving it one.
This is what happens when adoption is the metric and value goes unmeasured. With nothing better to measure, speed becomes the proxy. The team optimises for what it can see: how fast it produces the work it already knows how to produce.
But the question almost nobody asks is the only one that matters.
Why are we still producing this at all?
The monthly report was built for a world where pulling that data together took a person a week, so the business accepted a monthly view of performance.
The content calendar was built for a world where producing content was slow and expensive, so marketing planned weeks ahead to ration the output.
The nurture sequence was built for a world where nobody had time to understand and respond to every buyer signal individually.
The campaign dashboard was built for a world where the business needed visible proof that marketing was doing the work, even if the dashboard never explained whether the work was changing the commercial outcome.
And the MQL was built for a world where marketing could not judge a potential customer’s real intent at scale, so it batched volume into a proxy, stamped it “qualified,” and handed it to sales.
The number went up every quarter. The pipeline did not.
The team kept running on the wheel anyway, because hitting the number was the job.
A lot of what a marketing function produces is a workaround for an old constraint.
That is the real shift.
AI does not just reduce the cost of producing marketing work. It changes the constraint the operating model was built around.
If reporting is no longer slow, why is the business still waiting for a monthly readout?
If content is no longer scarce, why is the team still organising around a content calendar?
If buyer signals can be interpreted continuously, why is follow-up still built around static nurture logic?
If sales calls, customer objections and win-loss patterns can be structured into reusable intelligence, why is most of that knowledge still trapped in people’s heads?
The mistake is treating AI as a productivity layer on top of the old model, when the real opportunity is to redesign the model around a different constraint set.
The work should not simply get faster. The operating model must change.
That is why the faster horse is dangerous. Speed makes the old work feel newly justified, even when the business no longer needs it.
The team becomes more efficient at producing output the business may not need.
And because the output is higher, the function can look more productive while moving further away from the number.
That matters in a PE-backed business because activity is not the asset. A premium multiple is underwritten by a repeatable, legible growth system.
A buyer pays up because the growth story looks underwritten by that system: clearer ICP, stronger conversion, better revenue quality, lower waste, faster learning, more repeatable execution, and a commercial operating model that does not depend on heroic individuals.
So the real move is harder and quieter.
For every recurring thing marketing produces, ask three questions.
What outcome was this originally meant to create?
Is that outcome still commercially important?
And in an AI-native operating model, is this still the best way to create it?
Sometimes the answer is a sharper version of the same thing. A better board report. A more useful sales narrative. A clearer campaign brief. A faster feedback loop.
More often, the answer is a different thing entirely.
The monthly report becomes a live view of commercial leakage.
The content calendar becomes a market education system built around the buyer objections showing up in real time.
The nurture sequence becomes a signal-driven follow-up motion.
The campaign dashboard becomes a decision tool, not a justification tool.
The AI pod becomes embedded inside the GTM operating model, close enough to sales calls, customer objections, pipeline data, positioning and product truth to build something the business can actually use.
That is the difference between using AI to go faster and using it to change what the company does.
One produces a faster horse and a busier team. The other builds a growth system the next buyer can see and value.
A useful test is to walk through the marketing function’s recurring outputs and ask which ones still exist because of an old constraint.
If the answer is unclear, that tells you something. You may be looking at an operating-model problem before a team, model or market problem.
It is one of the first things I look for inside a marketing function.
You can often see the pattern in days.