WARC Study: AI Agents Are Turning Marketing Upside Down

For decades, marketing was defined in relatively simple terms. Brands had to attract attention, spark interest, build trust, and guide customers through the purchasing process as smoothly as possible. Search engines, comparison sites, retailers, and media outlets played a role along the way—but in the end, it was always a person who made the decision.

The End of the Traditional Customer Journey?

This fundamental principle is beginning to change.

That’s because a new intermediary is emerging between brands and consumers: AI agents. These digital assistants not only gather information but also compare offers, make recommendations, and—in the future—will be able to carry out entire transactions on their own. What is beginning today with ChatGPT, Gemini, or Perplexity is likely to evolve into a new form of digital commerce within a few years.

It is not the person who is disappearing from the process. But their purchasing decisions are increasingly being prepared, filtered, and, in some cases, delegated.

This new study examines precisely this shift «From Abundance to Agents», which WARC published in collaboration with the Omnicom agency PHD. It describes a paradigm shift that goes far beyond the use of generative AI in marketing. It addresses the question of how marketing works when machines become active participants in the purchasing process.

The real bottleneck isn't information—it's attention

The study begins with a simple observation: The digital world has long been producing more information, products, and content than people can process. The problem is no longer a lack of choice, but rather an oversupply.

AI agents are designed to solve exactly this problem: They handle repetitive tasks, analyze countless options simultaneously, create shortlists, compare prices, check reviews, and organize entire booking processes.

In the future, marketing will no longer be competing solely for human attention. It will be competing for relevance within a machine’s decision-making logic.

$3.35 trillion is a game-changer

The figures from the study show why this issue is much more than just a pipe dream.

As early as 2026, approximately $944 billion in consumer spending worldwide is expected to be supported by agent-based AI. Four years later, according to the forecast, this figure will rise to $3.35 trillion—nearly four times as much. By then, 3.8 percent of all global consumer spending would already be influenced or processed by AI agents.

What is noteworthy here is not so much the absolute number as the trend. After all, marketing rarely changes overnight. New touchpoints usually emerge gradually—until they suddenly become the norm. This is precisely where AI agents might be today.

Not every industry will be affected to the same extent

What is particularly interesting is that WARC and PHD are examining this shift on an industry-specific basis for the first time.

The markets most severely affected are those where people already have to spend a lot of time comparing, searching, and calculating.

These include, in particular, telecommunications, financial services, and travel and transportation. The study expects these sectors to see the fastest transition to so-called agent-to-agent transactions, in which AI systems independently compare and prepare offers.

The situation is different in sectors such as the automotive and healthcare industries. In those sectors, the final decision still rests with humans. While AI supports research and comparison, it does not replace humans due to complexity, regulatory requirements, or emotional factors.

Retail: When Retailers Become Interchangeable

The study is likely to be read with particular interest in the retail sector. That's because it describes a scenario that is likely to cause concern for many retailers.

When AI agents take over shopping in the future, they will compare retailers with a level of thoroughness that humans can hardly match. Delivery time, price, availability, and service will be weighed against one another in a matter of seconds.

For retailers, this means an uncomfortable truth: Those who simply sell the same products as everyone else run the risk of becoming interchangeable in the eyes of AI systems. Competition is therefore shifting from advertising messages to operational excellence, reliable supply chains, and clearly structured product data. At the same time, brand management remains crucial to ensure that consumers feed their preferences into the agent in the first place.

The media and publishing industries face a twofold challenge

The study also contains noteworthy findings for media companies.

In the future, recommendation engines and AI assistants will increasingly determine which sources are even visible.

This fundamentally changes the role of journalistic brands: Publishers must not only build trust with their readers but also become a trusted source for AI systems. Building their own relationships with users is becoming even more important because it ensures independence from external platforms.

For specialized media in particular, this is likely to mean that journalistic quality alone will no longer be sufficient in the future. Content must also be structured, machine-readable, and semantically unambiguous.

Four examples show how marketing is already responding today

What makes the study particularly compelling is its focus on companies that are already consistently aligning their marketing strategies with AI agents. The four examples come from completely different industries—yet they exhibit surprisingly similar patterns.

NatWest: Visibility Starts Today at the ATM

The British bank recognized early on that traditional search engine optimization would no longer be sufficient in the future. As more and more people are asking financial questions directly to AI systems, NatWest developed structured, semantically annotated datasets based on over 10,000 real customer conversations. The goal was no longer to appear as high as possible in Google search results, but rather to be selected by AI systems as a trustworthy source. The strategy paid off: visibility in AI-generated responses increased by over 500 percent, and the bank also opened about half a million new savings accounts.

Magalu: When the Brand Mascot Becomes a Salesperson

The Brazilian retail group Magalu took it a step further. Its virtual brand character «Lu,» which has been well-known for years, was evolved from a digital mascot into a full-fledged AI agent. Via WhatsApp, “Lu” advises customers in natural language, answers questions, recommends products, and guides them through the entire purchasing process until completion. Within just a few months, this led to millions of conversations, significantly higher conversion rates than through the traditional app, and revenue of several dozen million Brazilian reais. The real key to success, however, lay elsewhere: The focus wasn’t on the technology, but on a familiar brand personality that could suddenly actually talk to people.

Audi: Fewer Data Sheets, More Decision-Making Support

Audi, too, used AI not primarily for automation, but to simplify complex purchasing decisions. Instead of presenting prospective buyers with long lists of technical specifications, the manufacturer developed an AI-powered advisor that recommends the right vehicle based on simple questions about lifestyle. The salesperson thus evolved from a mere product seller into a guide through the decision-making process. The result was a significant increase in qualified leads and a conversion rate that more than doubled.

Dove: Authenticity Is Becoming the Currency for People and Machines

Perhaps the most surprising finding comes from Dove. The brand deliberately avoided perfectly staged advertising messages and instead placed unfiltered Reddit reviews of its hair care products at the center of its campaign—including critical voices. This radical transparency not only strengthened consumer trust but also significantly increased visibility in AI systems such as ChatGPT. The study thus impressively demonstrates that machines are increasingly guided by authentic community discussions. Therefore, anyone who wants to be credible must not only convince people but also generate the data on which AI bases its recommendations.

Marketing is becoming a discipline for two target audiences

However, perhaps the most important finding of the study is not that machines will influence purchasing decisions in the future.

Rather, it’s that, for the first time, marketing must cater to two completely different target audiences simultaneously. People want emotions, stories, experiences, and a sense of connection. Machines require structured data, unambiguous attributes, consistent information, and verifiable signals of trust.

Bringing these two worlds together is likely to become the central management challenge in the coming years.

The brand is becoming more important—not less important

Anyone who believes that AI agents make traditional brand management obsolete is likely misunderstanding the study.

The opposite seems to be true. Precisely because machines can compare information, strong brands are becoming increasingly important. They create preferences that consumers pass on to their agents. At the same time, distinctive brand values provide the guidance that AI systems can also take into account when making recommendations.

That is why the study concludes with a remarkably classic insight: Brands should not have to choose between people and machines. They must be relevant to both.

Structured data, machine-readable content, and semantic precision will be just as important in the future as creativity, storytelling, and emotional brand management. Those who optimize solely for algorithms lose sight of people. Those who appeal only to people risk being overlooked by machines.

Marketing isn't losing its most important target audience—it's gaining a second one. This is likely to be one of the biggest strategic challenges of the coming decade.