Monks Co-Founder and Chief AI Officer Wesley ter Haar expects advertising inside large language models to cannibalize generic search budgets as consumers shift product research and purchasing to AI agents. He told Marketing Report that the real transition is from being findable to being executable - a brand must be usable for whatever an agent does next.
The three layers of LLM advertising
Ter Haar described ChatGPT Ads as one component of a broader shift where paid placement, visibility through trusted sources and signals (generative engine optimization), and agent-ready product data converge. "The real step is from findable to executable: a brand must not only appear in the answer but be usable for whatever an agent does next," he said. "That is where we help our clients."
He advised brands in categories where expertise, personal preference and product comparison matter to develop a mature LLM and agent strategy. For advertisers whose audiences are barely active on these platforms or whose product information is not in order, he recommended holding off on substantial media budget shifts and focusing on those fundamentals first.
Measurement remains an open question. OpenAI currently reports impressions, clicks and conversions, but ter Haar pointed to a deeper issue: how much of the funnel takes place inside the LLM or is executed by agents from the LLM. "An attributed conversion is not yet demonstrable incremental sales," he said.
GEO, paid visibility, and the coming budget shift
Ter Haar drew a clear line between GEO and LLM advertising. GEO is the strategy for creating and optimizing the information, content and context that LLMs and agents use to judge a brand. LLM advertising adds a paid marketing message on top. Good GEO can reduce dependence on advertising, he said, but it does not make paid visibility redundant. Brands need to test where it still delivers extra value.
On pricing, ter Haar was blunt: "At the moment you are paying tuition and paying for the hypothesis that an ad inside an LLM environment has more impact. That added value still has to prove itself." A higher cost per click becomes defensible only when it demonstrably delivers more business.
His forecast for media mix disruption is specific. Initial spending will come from innovation and test budgets. If results hold and spending scales, he expects cannibalization to hit generic search advertising, affiliate marketing, and parts of retail media and social media. If consumers start running choices and purchases through LLMs and agents at scale, the focus shifts toward CX - "where the C stands for consumer as well as computer."
Trust, personalization, and the European market
On trust, ter Haar argued that advertising in LLMs does not have to be more irritating than a non-skippable pre-roll. What matters is clarity about what is paid for and the principle that an advertiser cannot buy the independent answer. He expects ad formats to evolve toward something more useful than the current interruption model.
Personalized advertising is not initially available in the European Economic Area under GDPR. Ter Haar sees the first opportunity in conversational context: someone states what they are looking for, what matters to them, and which trade-off they are making. "You can already connect to that very relevantly," he said. He expects people using a free product to prefer useful advertising over random advertising.
He also predicted interactive branded content inside LLMs - product pages that adapt to a question, configurators inside the conversation, or personal product demonstrations after a photo upload. He said paid services could emerge that let brands manage their official knowledge base for LLMs and agents, though buying a place in the independent answer remains a separate step that he expects platforms to keep walled off for now.
Competitive landscape: Anthropic, Google, Mistral
Ter Haar noted that not every AI company will make the same choice on advertising. Anthropic has opted for an ad-free Claude, earning from subscriptions and business contracts. Google is developing ad formats for its AI search experiences, though the standalone Gemini app could remain ad-free as a competitive differentiator. For Mistral, he sees more logic in paid professional use aligned with its Vibe positioning around work and coding. "To become a large advertising business you first have to build a lot of structural consumer usage," he said.
The Netherlands, he added, looks like a good test market: relatively small, digitally mature, and critical. Self-service is now available to Dutch advertisers, making hands-on testing more concrete. He declined to grade OpenAI's approach yet, preferring to revisit at the end of the year when more product iteration has occurred. For marketers building skills in this area, Generative AI and LLM Courses offer practical grounding in the technologies driving these ad formats.
Why this matters for marketing professionals
Ter Haar's core argument is that LLM advertising is the visible tip of a larger structural change: the purchase process moving inside conversations and agent-driven workflows. For marketers, the operational implication is that product data, pricing, availability and transaction processes must be machine-readable and agent-executable - not just discoverable. Brands that treat this as a search-ad extension will lag those that rebuild for agent compatibility. The budget pressure on generic search and affiliate channels is not hypothetical; ter Haar expects it to materialize as testing yields positive results. Marketing teams should audit whether their product information stacks are ready for an agent to act on, not just for an LLM to cite. For structured learning paths on these shifts, AI for Marketing Courses cover strategy through execution.
Your membership also unlocks:
