If you think AI will fix your marketing, you're barking up the wrong tree

Over the last year or so, I've sat through more AI demos than I can count. Most are genuinely impressive. A brief goes in and, within minutes, out comes a campaign idea, 3 customer segments, 5 social posts, a landing page, an email journey and perhaps a dashboard for good measure.

Work that would once have involved 6 people, 14 emails and a couple of "quick alignment calls" suddenly appears in minutes.

And inevitably, someone says, "This could save us so much time."

Probably. But what happens after the AI produces the work? Who decides whether the brief was right? Which customer data are we using? Who owns the campaign if brand, digital and sales disagree? And who owns the number at the end?

That is usually when the exciting AI conversation turns into a rather uncomfortable marketing conversation. And I think that may be one of the best things AI does for marketing.

It exposes what we have spent years learning to work around.

We have built marketing on invisible human glue

Marketing organisations have always survived on workarounds. Someone knows whom to call. Someone knows the CRM data is not quite right. Someone knows legal normally takes four days unless you call the right person. Someone also knows that the "official" customer journey looks very different from what customers actually do.

Capable people spend a surprising amount of time stitching all of this together. We rarely call it broken. We call it "the way things work here."

AI changes that.

AI is not particularly good at organisational folklore. You cannot scale an AI-enabled marketing function around instructions such as, "Priya usually approves this, unless it is Europe, in which case check with Raj." Humans can survive ambiguity. Machines tend to amplify it.

We are buying AI faster than we are redesigning marketing

Businesses are investing in AI at remarkable speed. Gartner's 2026 CMO research says AI already accounts for an average 15.3% of marketing budgets.

But only 30% of CMOs describe their AI readiness as mature, while 70% say their internal marketing processes are not mature enough to scale AI effectively. McKinsey has found something similar — AI usage among marketers is already widespread, but very few organisations are capturing value across complete, end-to-end workflows.

That gap matters. Using AI is easy. Redesigning the way marketing works is not. And I suspect we are about to discover the difference.

AI can make a bad process faster

Take content. If your positioning is fuzzy, AI can now produce much more fuzzy content than your team ever could — faster, cleaner and in twelve variations.

Or take data. If marketing, sales and customer service have different definitions of a "high-value customer", AI does not create a single customer truth. It simply gives you a more sophisticated way of analysing the disagreement.

Then there is workflow. If a campaign currently passes through 8 approvals, AI may reduce content creation from 3 days to 3 minutes. But the campaign still spends 11 days waiting for 8 approvals.

We have optimised the wrong bit.

This is why I get slightly wary when organisations begin their AI journey by asking, "Which tools should we buy?"

The more useful question is — how should this work happen now?

The real unit of transformation is the workflow

Once you stop looking at marketing as a collection of departments and start looking at it as a series of workflows, the gaps become obvious.

How does an insight become a campaign? How does a lead become a sales conversation? How does customer feedback get back into positioning? Where does judgement sit? Where does accountability sit? And where does work simply stop and wait?

That is where AI becomes transformational.

Not when it writes the email faster, but when you redesign the chain around what humans should do, what machines should do and, importantly, what probably should not be done at all. That last part gets very little attention.

AI has made producing things incredibly easy. Which means marketing teams may finally need to ask — do we actually need to produce all these things? Do we really need 12 campaign variations? 7 dashboards? 4 approval layers? A content calendar filled because Tuesday requires a post?

When creation becomes cheap, judgement becomes more valuable.

The CMO does not need to become the Chief Prompt Officer

Knowing what not to do may become one of the most important marketing capabilities. There is a temptation for marketing leaders to think staying relevant means learning every model, tool and new feature. Technical literacy matters — a CMO needs to understand what is possible.

But the job was never to operate the tools. The job is to decide what the business is trying to be famous for, where judgement must stay human, and which parts of the machine deserve to be rebuilt rather than accelerated.

AI will not fix your marketing. It will show you, very quickly and very publicly, where your marketing was never designed in the first place.