For the past few years, most conversations about large language models have focused on what users can do with them. We ask a questions, summarize a documents, generate an images, research some topics, compare products, write an email…
But as LLM platforms become a more regular part of how people discover information and make decisions, another question is becoming increasingly important for the advertising industry: what happens when these environments become media channels themselves?
The answer is unlikely to be as simple as placing traditional display ads next to chatbot responses.
LLMs introduce a fundamentally different interface between users, information, and commercial intent. They can understand a request, maintain context throughout a conversation, research alternatives, narrow choices, and increasingly assist users in taking action. This creates the foundations for a new advertising ecosystem where the valuable asset may not be a pageview or an impression, but the moment when intent becomes clear.
The LLM advertising market is therefore likely to develop differently from the web advertising market that preceded it. The question is no longer simply where an ad can be placed. It is where commercial value exists inside an AI-mediated journey, how advertisers can participate without damaging trust, and how that participation should be measured and monetized.
LLMs are becoming a new layer of digital discovery
Traditional digital advertising developed around destinations. Users visited search engines, publisher websites, social platforms, marketplaces and apps. Each destination created inventory and advertising technology evolved to connect that inventory with demand.
LLM interfaces disrupt this structure because they increasingly operate across destinations.
Imagine someone researching software for their company. Traditionally, that journey could involve a Google search, several publisher articles, product websites, comparison platforms, Reddit discussions and perhaps a few YouTube reviews.
With an LLM, much of that process can happen inside one conversation.
A user can explain the size of the company, budget, existing technology stack, required features, and concerns, then ask the model to compare several options. They can continue asking questions until the shortlist becomes increasingly specific.
For advertisers, that creates something extremely valuable: context that develops progressively around an actual need.
Digital advertising has spent decades attempting to infer what consumers want from fragmented signals. Search queries, browsing behavior, content categories, audience segments and commerce activity all help build an approximation of intent.
Conversational environments potentially bring advertising much closer to the moment when that intent is explicitly expressed.
That does not mean advertisers should have access to private conversations or that every prompt should become a targeting signal. Privacy and user expectations create obvious boundaries. But it does mean that LLM platforms are creating a type of commercial environment that does not fit neatly into existing definitions of search, social, display or commerce media.
Advertising cannot simply be inserted into the conversation
The most obvious monetization model for an LLM platform would be familiar advertising: sponsored placements alongside responses.
But conversational interfaces create a much more sensitive trust problem than traditional webpages.
When a search engine displays a clearly labeled sponsored result above organic results, users generally understand the distinction. When an AI assistant recommends one product over another inside a natural-language answer, the boundary can become less obvious.
Was the product recommended because the model considered it the best option? Was it included because the brand paid for visibility? Did sponsorship influence its position? Would the answer have been different without commercial involvement?
These questions matter because users interact with AI assistants differently from conventional advertising environments. The interface itself is designed to provide answers and assistance. If commercial influence becomes difficult to distinguish from the model’s judgment, the usefulness of the product can quickly come into conflict with its monetization model.
This is why transparency will probably become one of the defining principles of LLM advertising.
Sponsored participation needs to be identifiable as sponsored participation. The commercial relationship cannot quietly disappear inside an apparently neutral recommendation.
For advertisers, this also means the objective should not be to force a brand into every relevant conversation. It should be to participate when the brand genuinely helps answer the user’s need.
The unit of advertising value could change
Much of digital advertising still revolves around the impression. A page loads, an opportunity becomes available and the market determines how much that opportunity is worth. Conversational AI makes that unit less obvious.
A user could spend ten minutes interacting with an LLM without generating ten conventional pages. Yet during that conversation, the system may learn that the person is planning a trip to Berlin next month, needs a hotel near a specific location, has a defined budget, requires parking, and prefers flexible cancellation.
From a commercial perspective, this could be substantially more meaningful than ten unrelated pageviews.
The advertising opportunity may therefore become less about where the user is and more about what the user is trying to accomplish.
This creates several possible monetization models.
Platforms could sell clearly labeled sponsored recommendations or placements connected to relevant queries. Commerce-oriented conversations could generate affiliate or referral revenue when users move from research to purchase. Brands could pay for qualified leads or completed actions rather than exposure. Some LLM environments could develop auction-based systems where advertisers compete for participation in commercially relevant moments.
Subscription and advertising models may also coexist. Users might pay for premium AI functionality while commercial integrations remain available in free products or specific shopping and discovery experiences.
The important point is that LLM monetization does not have to reproduce the economics of display advertising. In many cases, cost per action, referral, lead, recommendation opportunity, or commercial outcome may be more natural than cost per thousand impressions.
Search advertising provides a useful comparison, but only to a point
Search is perhaps the closest existing advertising model because both environments begin with expressed intent.
Someone searching for “best CRM for a 50-person company” is already providing a valuable commercial signal. Search advertising works because advertisers can respond to that intent at the moment it appears.
LLMs can potentially take this considerably further. A search query is often a snapshot but the conversation is cumulative.
The user may begin by asking about CRMs, then explain that the company operates in three countries, requires Salesforce integration, has a limited implementation team and wants to remain below a specific annual budget.
Every additional interaction clarifies the original intent. This creates the possibility of advertising that responds not simply to a keyword but to a much richer understanding of the problem being solved.
However, it also makes responsible advertising more complicated. The richer the context, the more carefully platforms must determine which information can legitimately be used commercially. The technical ability to understand a conversation does not automatically create permission to monetize everything contained within it.
The companies that build sustainable LLM advertising products will need to treat that distinction seriously.
Commerce may be where the model becomes clearest
Some of the strongest opportunities for LLM monetization are likely to emerge around commerce.
Product discovery already fits naturally into conversational interfaces. Instead of filtering hundreds of products manually, users can describe exactly what they need: “I need noise-cancelling headphones under $300 that are comfortable for long flights and work well with both my laptop and Android phone.”
An AI assistant can translate that request into criteria, compare products, explain trade-offs and narrow the selection.
At that point, the distance between discovery and transaction becomes very small.
Brands, retailers, marketplaces, and affiliate networks all have potential roles in this journey. Sponsored products could compete for clearly identified visibility. Retailers could provide current prices and availability. Brands could supply structured product information. Platforms could receive referral revenue when a recommendation results in a purchase.
This is where LLM advertising begins to overlap with retail media and commerce media. The advantage is not simply that AI can show products. It is that AI can understand why a particular product fits a particular request. If executed well, commercial relevance could therefore improve rather than interrupt the user experience.
Brands will need to optimize for machines as well as humans
The emergence of LLM-driven discovery also changes what it means for a brand to be visible.
Traditional digital marketing has focused heavily on making information accessible to search engines and attractive to human users. In an LLM ecosystem, another audience becomes increasingly important: systems attempting to understand, compare and recommend products.
A beautifully designed product page may work well for a person while providing poor structured information for an AI system.
Brands may therefore need to think more seriously about the accessibility and quality of their product data. Specifications, pricing, availability, reviews, differentiators, documentation, location information and other commercial details need to be accurate and understandable.
This creates an interesting shift in advertising strategy.
Brands will still compete for attention, but they may increasingly compete for machine understanding and recommendation eligibility as well.
Being visible to an AI assistant does not automatically mean paying for advertising. Organic authority, reliable information, strong reviews, publisher coverage, and clear product data can all influence discovery.
Advertising becomes another layer on top of that foundation rather than a replacement for it.
Publishers have a complicated role in this ecosystem
LLMs need information and much of that information originates outside the platforms themselves.
Publishers, reviewers, specialist websites, forums, researchers, and other open-web sources contribute to the knowledge that makes AI-assisted discovery useful.
This creates a difficult economic question. If an LLM uses publisher content to help a user make a commercial decision but the user never visits the original website, where does the publisher participate in the value chain?
This is especially important for publishers whose business models depend heavily on advertising impressions and referral traffic.
The LLM advertising ecosystem cannot be considered independently from this issue. If AI platforms capture both the user’s attention and the advertising opportunity while external publishers provide much of the underlying information, the economics of content creation become increasingly strained.
Licensing agreements, paid content access, referral models, attribution mechanisms, revenue-sharing arrangements and other forms of compensation may therefore become an important part of the emerging market.
The long-term sustainability of LLM advertising will depend not only on how AI platforms monetize users but also on how value is distributed among the organizations supplying the information that powers those interactions.
What does this mean for programmatic?
The emergence of LLM advertising does not necessarily make programmatic infrastructure less relevant. In many ways, it creates another problem that automated advertising systems are well suited to solve.
If millions of conversational interactions create commercially relevant opportunities, advertisers will need scalable ways to evaluate them. Platforms will need mechanisms for matching appropriate demand with appropriate contexts. Pricing will need to reflect competition and value. Campaign objectives, budgets, frequency, quality, measurement, and brand suitability will still matter.
Those are familiar programmatic problems. What changes is the signal being evaluated. Instead of asking only whether a specific impression is valuable, advertising systems may increasingly evaluate whether a commercial moment, recommendation opportunity, or expressed intent is valuable.
DSPs could eventually incorporate new forms of conversational or agent-mediated inventory. Supply-side platforms may help package and standardize emerging opportunities. Measurement providers will need to determine how exposure inside AI environments relates to downstream actions.
And if multiple LLM platforms create their own advertising systems independently, interoperability will become increasingly important. Advertisers will not want entirely different infrastructure, definitions, reporting, and buying processes for every conversational platform. The market will eventually need standards.
Beyond chatbots
It is easy to think about LLMs as another digital product category. But their potential impact on advertising is larger because they can become an interface connecting many existing categories at once.
Search, commerce, publishers, apps, services, travel, local discovery, product research and eventually agent-driven transactions can all pass through conversational systems.
Where those journeys contain commercial intent, advertising opportunities will follow. The resulting ecosystem will probably look very different from today’s display market. It may combine sponsored recommendations, commerce integrations, referrals, performance-based models, programmatic auctions, subscriptions, and entirely new formats that have not yet been standardized.
What will remain familiar is the underlying challenge: connecting advertisers with relevant opportunities without compromising the experience that created those opportunities in the first place.
The LLM advertising market is still at an early stage. There is no settled format, dominant buying model or universal measurement framework yet. That uncertainty is precisely what makes the space important to watch. The next major advertising channel may not look like an advertising channel at all. It may simply look like a conversation.
