The CMO’s new brief: governing your brand in AI answers

Every CMO I speak to can recite the channels they are responsible for. The website. Paid. Earned. Social. The sales narrative. Each one has an owner, a budget, a way of knowing whether it is on message. That map is how a marketing leader has always answered the only question that really matters at board level: are we being perceived the way we intend?

There is now a channel on that map that no one was hired to run. When a customer asks ChatGPT, Claude, Gemini or Perplexity about your category, an answer comes back instantly and confidently, and it carries your brand inside it. That answer is shaping the purchase. It sits on no one’s org chart. And for most organisations, no one can say what it actually said.

The brief quietly grew a new line

For decades the CMO’s mandate was to author the brand and defend it across the channels the company controlled. The words, the framing, the proof, all of it ran through surfaces you owned. The discipline was consistency.

The AI answer breaks that model. It is a surface that speaks for your brand to a buyer who trusts it, and you did not write a word of it. You cannot brief it the way you brief an agency. You cannot approve it the way you approve a campaign. It assembles its answer from whatever the web has said about you, your site, yes, but also a retailer’s outdated copy, a three-year-old press cycle, a competitor’s framing, a Reddit thread that happened to get cited fifty times.

So a new line has appeared on the brief, whether or not anyone wrote it down. Not make the brand visible in AI: visibility is the easy part, and plenty of tools already count it. The line is make sure the brand is represented correctly in AI. That is not measurement. That is governance.

Governance is a different job from monitoring

Most teams reach first for a monitoring tool, because that is the reflex the last decade trained. Count the mentions, plot the trendline, compare against the competitor next door. Useful, and not the job.

Monitoring tells you that you were named four hundred times last week. Governance answers the questions that come immediately after, the ones a CMO actually gets asked in the room: Did the model frame us as the premium house or the mass-market one? Did it pair us with our real competitors or the wrong ones? Did it carry our sustainability claim or quietly drop it? When it got something wrong, where did the wrong story come from, and who on my team can fix the surface it came from?

None of those are counting questions. Every one of them is a question about meaning, about accountability, about action. A dashboard that goes up and to the right cannot answer a single one of them, any more than a reach number ever told you whether the ad was on brand.

What governing the AI answer actually looks like

Governing a channel means the same three things it has always meant, applied to a surface that happens to be a model.

First, you read what is being said, not how often, but what. You interpret how each major model actually represents the brand: the category it places you in, the claims it repeats, the competitors it hands you, the story it tells when a buyer asks the question that leads to a sale.

Second, you trace each gap to its source. A misrepresentation is not weather; it has a cause. The model learned the wrong category from somewhere specific, and that somewhere is usually a public signal you can influence: a page, a listing, a stale description, a piece of coverage.

Third, you route the fix to whoever owns that surface, and you verify it moved. The retailer copy gets corrected. The page ships. The next read confirms the model’s interpretation has begun to shift. That loop - read, trace, correct, verify - is what turns a channel you can only watch into one you can actually run.

The line that now belongs on the agenda

The old question, do people trust what AI says about us, is settled. They do, and they act on it. The question that now belongs on the executive agenda is the governance one: when AI describes our brand, is it describing us the way we intend, and if not, who is fixing it, and how do we know it worked?

That question is not going to un-ask itself. The share of the customer journey running through AI answers only grows from here, and every quarter it goes ungoverned is a quarter the market forms its impression of you from sources you never chose. The CMOs who treat the AI answer as a channel to govern (with an owner, a workflow, and a way of knowing it is on message) will be the ones who can still say, honestly, that they know how their brand is perceived.

The brief already changed. The only open question is who picks up the new line.