This is what IT does for us. The most expensive sentence in marketing

Last week I sat down with the CMO of a major kitchen and home appliances brand. He was candidly unsure whether he actually needed AI enablement. “Copilot is in place,” he said. “I’ve built the first agents myself and I use them every day. And IT is building a company-wide chatbot just for internal use.”

None of that is wrong. All of it is more than most marketing organizations can show.

Still, the conversation stayed with me. Not because something was missing from his list, but because everything on it belonged to the same category: each item was a question of provisioning. Not one was a question of value creation.

I’ve had similar conversations several times in recent months. Increasingly I wonder whether marketing has adopted the right mindset.

AI is not software you “roll out”

You roll out an ERP system. It has a purpose, defined processes, and an end point. After that it runs.

AI doesn’t work like that. It’s an enablement technology, and without process competence it remains useless. It changes how marketing creates value: in content processes, search, campaign management, reporting, and increasingly in the roles themselves. So the question isn’t how to introduce AI technically. The question is how marketing’s value creation changes as a result. Only marketing can answer that.

That is not a critique of IT. On the contrary: IT is doing its job—security, governance, infrastructure, stable operations. IT builds the foundation that makes efficiency possible. Marketing has a different task: generate growth, develop demand, create competitive advantage.

If you hand responsibility for AI to IT, you usually get exactly what IT is optimized for. And that is efficiency.

Efficiency is a path, never the goal

Efficiency now happens almost by itself. Content gets produced faster, reports run automatically, research takes a fraction of the time it used to. That’s why efficiency has become the dominant metric for many AI initiatives.

That makes sense but it’s incomplete. Efficiency only says the same task costs less effort. The economic value that follows is decided afterward. And that decision isn’t always made by marketing.

US vendor Clarecast reported an uncomfortable finding: over 1,300 companies already show all four signals of AI-driven downsizing long before a single role is officially removed. Nearly 10,000 more are at risk. One detail bothers me most: affected teams ran about 17 months below their expected headcount curve before anything was announced. The decision had already been made while the departments continued to talk about productivity gains.

• Over 1,300 companies – already show all four signals of AI-driven downsizing
• Four signals – indicators of AI-driven shrinkage in the companies studied
• Nearly 10,000 more – are considered at risk
• About 17 months – affected teams ran below the expected headcount curve

Purna Virji, formerly at Microsoft and LinkedIn, therefore calls saved time the ultimate AI vanity metric. Her point lands: if you report AI success in hours saved, you tell a productivity story. And that makes the most persuasive case for doing the same work with fewer people going forward.

Marketing talks productivity. Finance reads that as cost-savings potential. Same number, different interpretation. Whoever provides the number but not the interpretation has surrendered it.

What IT can’t answer

The dividing line is clearer than the debate often suggests. Which tool reads competitors’ prices can and should be answered by IT. That price monitoring can be taken to a completely different level with AI is not the first reflex of most IT departments. That perspective has to come from the business—or it won’t come at all.

Three examples I see most often in projects.

Which content should we even produce?

For years, content production was limited by editorial hours. Topics had to be prioritized because resources were scarce. That constraint is gone. Producing content quickly is a no-brainer now. The more interesting question is which content should be created at all and what contribution it makes to value creation. That isn’t a technical question; it’s a prioritization decision with clear outcome accountability.

What happens to visibility if users stop clicking?

A whole practice has emerged around Generative Engine Optimization. Which content gets quoted, which formats work, which technical requirements apply—all valid questions. The real shift isn’t a new optimization discipline: your customers’ search behavior is changing. Information is no longer consumed on single pages but delivered directly by AI systems. Reach, distribution, and the entire logic of digital visibility shift. So you don’t only need to adjust SEO tactics. You need to decide what role visibility will play in your value creation going forward.

What are people for when agents take over workflow steps?

Agents prepare briefs, analyze audiences, generate ad variants, optimize budgets. That changes steps in the workflow. More important, it changes people’s roles. Competitive advantage will increasingly come less from executing operational tasks efficiently and more from making better decisions, integrating different information sources, and turning those insights into new offers and customer experiences.

All three examples show the same thing: AI doesn’t change single marketing disciplines. It changes the logic by which marketing creates value.

1. Content prioritization instead of production limits
   Editorial time used to cap what could be produced. With AI that cap disappears. What matters now is which content should be produced and how it contributes to value creation.
2. Visibility beyond traditional SEO
   Search behavior shifts toward AI-delivered answers, moving reach and distribution. The question is not only optimization but what role visibility plays in your value creation.
3. People as decision-makers, agents as operators
   Agents handle operational steps like briefs, variants, and budget optimization. People create advantage by making better decisions and combining information into new offers and experiences.

That’s why it’s too narrow to discuss content, search, and campaign management separately. The crucial capability is to think about these disciplines end-to-end: from first market observation through content and distribution and back into the organization.

And that brings us to the part that is least delegable. Marketing’s task has always been to understand customers and serve their needs as well as possible. That competence doesn’t shrink; it becomes the bottleneck. Someone must teach that knowledge to the agents (see Context Engineering). Without this context no agent consistently produces usable results. In the companies we support, that is regularly the tipping point: the technology exists, the knowledge does not.

What comes after efficiency?

Efficiency gains are appearing everywhere. Models get better, agents take on complex tasks, manual processes disappear. The key question is what you do with those gains.

If your answer is to do the same work with fewer resources, AI remains a rationalization project. If your answer is to use the same team to develop more, enter new markets, or build different customer experiences, you get a very different outcome.

A concrete first step that takes an hour and is fully reversible: list your three largest AI initiatives and write down which business outcome each one enables. Not how much time it saves. Which outcome. Where you cannot answer that, you face a provisioning question, not a value-creation question. That’s not yet a problem. It becomes one as soon as someone else in the company formulates the answer for you.

Frequently asked questions about AI, efficiency, and marketing accountability (FAQ)

Why isn’t it enough to introduce AI technically?

Because AI is an enablement technology that only delivers impact when supported by processes and target outcomes in the business. Pure provisioning produces at best efficiency, not an answer to how marketing creates value with AI. Only marketing can provide that answer.

How do I measure AI success beyond hours saved?

Measure business outcomes, not hours. Define the result an initiative should enable and link the work to clear outcome accountability. Time savings are a means, not the goal.

What role does IT have compared with marketing?

IT ensures security, governance, infrastructure, and stable operations. Marketing is accountable for growth, demand, and differentiation. If you hand full responsibility to IT, you will primarily get efficiency—not necessarily impact.

What does the change in search mean for my visibility?

Customers increasingly receive answers directly from AI systems without visiting individual pages. That changes reach, distribution, and the logic of visibility. The key question is what role visibility should play in your future value creation.

Where do I start if we’ve focused on efficiency so far?

List your largest AI initiatives and note the intended business outcome instead of hours saved. Where that mapping is missing, you’re dealing with provisioning. That creates clarity and shifts the conversation to impact.

Back to the conversation at the start. The CMO’s question—whether he needed enablement—was honestly asked. It was just addressed to the wrong team. He has AI. What he lacks is an answer to what his marketing will do differently with it.

Technology amplifies your organization. It does not improve it.

If you give away responsibility, you get efficiency. If you keep it, you can use the same technology for growth, innovation, and new value creation. Now is the moment to make that leap because efficiency gains are only just becoming visible and no one has yet decided who will own them.

Who in your company answers the question, “what comes after efficiency”? If you can’t think of anyone, that answer may already have been made.