Contentful x Atlantic Insights: Five Questions Every Marketing Leader Needs to Answer About AI

The next era of marketing will be defined by two kinds of judgment: the judgment humans retain when working with AI and the judgments machines make about brands.

Our collaboration with Atlantic Insights last year surveyed the landscape of how machines were helping humans in marketing. This year, the conversation has evolved.

Per our latest research, 91% of marketing leaders agree that the future of marketing increasingly depends on persuading the systems that influence people.

That creates a different challenge for CMOs. And the question is no longer How should we adopt AI?

These are the new questions every marketing leader needs to be thinking about now.

1. Autonomy: What can our agents do without us?

We've spent a lot of time asking what agents can do. Now we need to decide what we're comfortable letting them do on their own.

Agents are already working across an average of 3.6 marketing functions per organization. But there is little consensus around how much autonomy they should have. Forty-three percent of teams would let agents act across most marketing tasks with periodic review. Thirty-four percent limit them to low-stakes decisions. And 21% require human approval before any action is executed.

There isn't one universally right answer. The risk of an agent sending an off-brand customer communication is different than tagging a piece of content incorrectly.

What matters is being intentional about the decision boundary — the line between what an agent can do, what a person reviews, and what a person must decide.

There's a cost to getting that boundary wrong in either direction. Give agents too much freedom, and you introduce unnecessary risk. Require a human to approve everything, and you create an approval tax. You’ve invested in automation for speed and efficiency, only to have the work sit idle while waiting for someone to sign off.

The goal isn't maximum autonomy. It's putting human judgment where it adds the most value.

2. Judgment: What has to remain human?

If agents take on more execution, what do we deliberately keep human?

Marketers already have strong instincts here. Creative ideas and strategic decisions are the areas where they trust AI least, at 44% each, followed closely by brand voice at 43%.

But there's another side to this we need to consider: 25% of organizations have reduced or paused entry-level hiring because of AI. And much of the work we're automating is exactly the work through which generations of marketers learned their craft.

We learn judgment by doing, and that creates lived experience. As an entry-level marketer, writing a bad first draft, watching a campaign underperform, or understanding why a manager changed a headline is all part of the learning process. Eventually, you develop an instinct for what is right for the brand and the business.

If we remove that work without replacing the learning that came with it, we create judgment debt: a future shortage of distinctly human skills that increase in value as AI handles more execution. And we won’t be able to hire our way out of it down the line.

As leaders, we can't just ask what work AI can take off people's plates. We have to ask what people were learning by doing that work, and how we're going to teach it differently.

3. Legibility: Can the agentic web understand and verify our brand?

For years, marketers have worked to make our brands compelling to people. Now we have another audience to consider: machines. And this is where the instinct to simply create more content can lead us in the wrong direction. At Contentful, we call this a content collapse. 

Eighty-five percent of marketing leaders believe AI summarization will make most brands sound the same. At the same time, 95% believe brands with strong, well-codified identities will widen their lead.

The difference isn't volume. It's signal integrity. AI systems don't form an understanding of your company from your homepage alone. They're drawing from product information, structured data, press coverage, reviews, citations, community conversations, and countless other signals.

Do those sources reinforce one another? Are your claims supported by evidence? Is your product described consistently? Can a machine understand what actually makes you different from your closest competitor?

When those signals conflict, we don't get to choose which version wins. The AI system does. So AEO is much bigger than optimizing a few pages for answer engines; it’s about creating a clear, consistent, verifiable account of your brand that can hold up wherever machines go looking for it. 

4. Measurement: What do AI systems say about our brand, and how is that changing?

Once you recognize AI as an audience, measurement has to evolve too.

Eighty-three percent of marketing leaders say showing up accurately in AI systems is a top or high priority for the next 12 months. Fifty-five percent already regularly measure how AI systems describe, summarize, or recommend their brand.

That's progress. But a snapshot only tells you what happened today. Models change. Sources change. Competitors change. Your own content changes. And over time, all of those inputs can gradually alter how AI systems understand, compare, and recommend your brand.

We call this representation drift.

The important question isn't: What does AI say about our brand? It's: How has that answer changed? Why did it change? And what can we do about it?

This is an area we've been thinking deeply about at Contentful. It's also part of why we built Palmata — to help organizations move beyond simply checking their AI visibility to understanding their broader AI reputation, the sources and signals influencing it, and where they have opportunities to improve.

Ultimately, managing AI reputation needs to become a discipline with a regular cadence, not an occasional spot check.

5. Ownership: Who is accountable for keeping the system aligned?

Today, marketing holds primary responsibility for brand visibility in AI systems at 45% of organizations we surveyed. Elsewhere, the work sits with AI and data teams, IT, engineering, or other members of the C-suite. Six percent still have no clear owner at all.

I understand why. No single function controls everything shaping how AI understands a brand.

Marketing owns the brand and much of the content. Product owns product truth. Data and technology control critical systems and signals. Communications influences third-party authority. The list goes on.

The work has to be cross-functional. But accountability should sit with the CMO. If AI systems increasingly shape how customers discover, understand, and choose your brand, how your company is represented in those systems is a marketing issue, even when many of the inputs sit outside marketing. Marketing already owns how the market understands the company. AI-mediated understanding is an extension of that mandate.

Shared responsibility is inevitable. Shared accountability can't be. Organizations need a clear accountability map: one leader accountable for the outcome, with named owners across the business responsible for the inputs they control.

The framework matters more than having perfect answers

One essential callout: These aren't questions you answer once.

Models will change. Agents will become more capable. Customer behavior will evolve. The right decision boundary today may not be the right one a year from now. That's why the framework in our report goes one step further.

For each of these questions, we look at four things: the decision that needs to be made, the risk of putting it off, the operating standard to work toward, and the cadence for revisiting it.

There are a few principles I think every marketing organization can start applying now:

• Write down your boundaries. Don't leave agent autonomy to individual interpretation. Define where agents can act, where humans review, and where humans decide.

• Protect the development of judgment. When you automate work, identify what people learned from doing it, and intentionally create another way to build that experience.

• Strengthen your signals before creating more of them. Establish governed sources for your brand and product truth, then make sure those facts remain clear and consistent wherever AI systems look.

• Measure change, not moments. Establish a baseline for how AI systems represent your brand, and revisit it consistently enough to spot meaningful drift.

• Put one person at the top of the accountability pyramid. The work can (and should) span functions. The outcome still needs an owner.

The insights from this research help paint a picture of what AI maturity in marketing really looks like. It isn’t about how many agents we deploy, how many workflows we automate, or how much content we can create. Organizations have to strike a balance between where AI adds value and where humans add value — giving machines greater freedom over the work they do well while creating more space for people to exercise creativity, judgment, empathy, and strategic thinking. AI doesn’t diminish the human role in marketing. It makes it more consequential. This is the opportunity for marketing leaders.

Take a tour

Chat with our team

Your experiences should wow, not wait. Let's talk.