A New Divide Is Emerging in AI Marketing: Brands Must Not Only Be Recommended, But Also Be Ready to Talk With Consumers

When I saw OpenAI release Sponsored Agents last week, my first reaction wasn't "ChatGPT is getting into advertising too."

What really made me stop and look was that a marketing motion I've observed for twenty years is starting to change: after users click an ad, where do they go?

According to information OpenAI published on September 16, after users see a relevant ad, they can choose to enter a clearly labeled Sponsored Agent conversation, continue asking about the brand's products or services, and then enter the brand website once they're ready. Currently, Sponsored Agents is still a limited alpha for select US advertisers — not all advertisers can use it.

So at this stage, it's too early to understand it as a mature ChatGPT Ads placement channel. But if we see it as a new interaction pattern emerging in AI marketing, I think it's well worth marketing teams' attention.

Why Does Sponsored Agents Deserve to Be Reunderstood From "After the Click"?

I started doing SEO at a Hong Kong digital marketing company in 2005. Back then, we were still dealing with Yahoo, portals, and early search engines. Later, Google became an important entry point for many businesses to acquire search traffic, and then came social media, short video, and today's AI search. I've basically experienced three generations of change in search and digital customer acquisition environments.

Interestingly, over the past twenty years the traffic entry points have kept changing, but what happens after an ad click hasn't actually changed as much as you'd think.

In the Yahoo and portal era, users entered the website after clicking; once Google search ads matured, brands started seriously designing Landing Pages; in the social ad era, entry points became feeds, video, and information streams, but ads still often brought consumers to pages the brand itself designed.

Landing Page has always been the brand's own territory.

Brands can decide what selling points to show, what to explain first, where to put the button, and what the user does next. Ads find the people, and the page handles the subsequent explanation and conversion — this division of labor has lasted for many years.

Where Are AI Ads and Traditional Search Ads Really Different?

One important signal Sponsored Agents gives me is that between the ad and the Landing Page, a layer of "conversation" may start to appear.

After users see an ad, they don't necessarily leave the AI environment right away. They can continue asking: Is this product suitable for my situation? What's the difference between the two models? Will the size fit? How many people can sit in it? How should the material be maintained? What OpenAI showed when announcing Sponsored Agents was exactly this kind of continuous follow-up scenario.

This means the role ads play may no longer be just to attract a single click. The conversation itself may also start to take on some of the explanation and comparison work that used to be handled by Landing Pages, product detail pages, or even customer service.

The click may no longer be the end of persuasion, but the beginning of a question.

But I don't think this means Landing Pages will disappear. At least from the current design of Sponsored Agents, consumers can still go to the brand website after the conversation. The more noteworthy change is that the single path of "ad → page" has the opportunity to become "ad → conversation → page or next step."

From Web Pages to AI Conversations, What Has Changed in Brand Marketing Assets?

If we put this back into the changes of the past twenty years, it becomes easier to see clearly.

This isn't a matter of one generation of channels replacing another.

Websites still need to handle transactions, brand experience, complete product information, service descriptions, and deep verification. SEO won't lose its value just because AI search has appeared. What's really loosening is the once-natural default premise: after consumers click in, the brand website mainly handles the persuasion task that follows.

Now, some of the explanation may happen earlier, inside the AI conversation.

Before consumers have even opened your product page, they're already asking about differences, use cases, limitations — even comparing your product with other options. At this point, whether the brand can enter the conversation, and whether AI can explain the brand accurately, becomes the new marketing question.

Why Does AI Conversation Push GEO From "Being Recommended" Toward "Being Able to Answer"?

This is also why I believe the relationship between Sponsored Agents and GEO is deeper than "ChatGPT has one more ad format."

When many companies first start understanding GEO, their first question is usually: "When users ask related questions, does AI mention my brand?"

Taking it a step further, they start asking: "Why is AI willing to cite me and recommend me?"

But when consumers start continuous follow-ups inside the AI environment, a third question appears: "When they keep asking, do I have enough reliable information for AI to answer with?"

For example: product specifications, target users, use cases, pricing policies, model differences, after-sales conditions, service areas — and when a product is actually not suitable. This information, once scattered across product pages, FAQ, help centers, news content, even sales materials, can all become the knowledge foundation for AI product discovery and AI recommendation.

At this point, GEO is no longer just about earning a single brand mention.

When my team and I proposed AIPO, we consistently understood website optimization, content planning, and platform distribution within the same system — and this is why. Companies need not only to add public information, but also to go back and check existing web pages: Are the product facts clearly stated? What are the relationships between entities? Do the FAQ answer the questions consumers actually ask? Do structured data and page body content conflict?

Brand knowledge assets must not only be findable, but also continuously explainable.

Will AI Agents Replace Landing Pages?

A few days ago, when we were talking about this change, a friend doing overseas growth asked me: "So does that mean we don't need to build Landing Pages anymore?"

I said: "Actually, I think websites will become more important — they just won't carry all the explanation tasks anymore."

He paused for half a second: "What do you mean?"

I said: "Before, pages were mainly written for people to read. From now on, some of that content also needs to be considered for whether AI can accurately understand it."

This concern is reasonable. But drawing the conclusion now that "AI Agents will replace Landing Pages," I think, is too early.

From the design of Sponsored Agents as currently published by OpenAI, the brand website remains one of the entry points for consumers' subsequent actions. AI conversations are better suited for immediate needs explanation, clarifying questions, and helping with comparisons; websites continue to handle complete information verification, brand experience, transactions, and more complex service flows.

So I'd rather understand it as Page + Agent, not Page vs. Agent.

The same goes for SEO and GEO. GEO doesn't replace SEO — it adds a new layer of requirements on top of existing search assets, oriented toward AI understanding, citation, recommendation, and conversation.

What Are "Conversational Brand Assets"?

I call them "conversational brand assets": the brand's product facts, differentiators, evidence, rules, and use cases not only exist on web pages, but can also be accurately understood by AI and remain consistent when consumers ask follow-up questions.

What the next stage of AI marketing adds is not just a new entry point, but a new way of brand expression: moving from showing consumers to being able to respond to consumers' questions.

These assets include at least four types.

• Fact assets: product parameters, pricing, applicable scope, service policies, inventory status, and other basic facts consumers need to confirm.
• Evidence assets: product descriptions, research materials, case studies, and third-party information that supports the brand's claims.
• Relationship assets: what's the difference between product A and product B, and which solutions suit different users, budgets, and use cases.
• Boundary assets: when something doesn't apply, what limitations exist, which questions can't be simply answered, and which need further human confirmation.

This is also where Conversational Commerce and GEO begin to intersect. The former focuses on how consumers complete discovery, comparison, and decisions through conversation; the latter increasingly needs to solve a foundational question: where does the information supporting these conversations come from, is it complete, and is it consistent?

A Gartner survey of 322 US consumers conducted in January 2026 also offers a signal worth watching. The survey shows that consumers are more accepting of AI helping narrow down product choices than of letting AI directly make purchase decisions for them. For example, in the household goods scenario, 31% of respondents were willing to let AI help narrow down choices; Gartner also noted that among the lower-risk categories covered in its survey, the highest proportion willing to let AI directly make purchase decisions was 11%.

What I focus on more isn't any single percentage, but this gap.

At this stage, for marketing teams, "helping consumers explain, compare, and narrow down choices" may be more worth studying than imagining an Agent that fully purchases on consumers' behalf. Whether brand knowledge is clear directly affects whether such conversations have value.

What Should Brands Prepare Now for AI Conversations?

So, if you're a CMO or marketing director, I wouldn't suggest focusing all your attention right now on "when can we run Sponsored Agents." It's still in a limited testing phase, and many companies have no urgent need to build complex plans around an ad format that isn't fully open yet.

But there are three things you can check right now.

1. Can AI today accurately explain "what we sell and who it suits"?
2. If users follow up for three rounds in a row, is the information your company has made public enough to support consistent answers?
3. Between your website, product pages, FAQ, and third-party platforms, are there obvious factual gaps or conflicts?

These three questions can actually already be folded back into GEO's daily work.

Website optimization isn't just about adjusting keywords — it's also about clearly stating product facts and entity relationships; content planning and platform distribution isn't just about increasing article volume, but filling in the real decision questions users have; performance monitoring shouldn't only look at whether the brand name appears, but can also observe whether ChatGPT, Gemini, Google AI Mode, and Perplexity explain the brand and products consistently.

Then, through data analysis, identify high-frequency errors, information gaps, and competitive comparison scenarios — and in turn decide what the next round of website and content should fill in.

This is also a relationship I've come to value more and more within AIPO's five rings: website optimization, content planning and platform distribution, performance monitoring, and data analysis are not four independent things — they should continuously correct each other. The remaining ring is to keep adjusting strategy based on these results, so that brand knowledge assets gradually adapt to the new search and conversation environment.

What Should Different Marketing Leaders Focus On Right Now?

If you're a CMO / marketing director, I'd suggest not rushing to discuss "whether to build a brand Agent." First check whether your company's current digital assets are sufficient to support continuous consumer follow-ups around your products.

If you're an overseas brand lead, I'd suggest first pulling out 20 questions consumers commonly ask before purchasing, and observing in ChatGPT, Gemini, Google AI Mode, and Perplexity how your brand information is explained. The point isn't simply counting how many times you appear, but finding factual gaps, misunderstandings, and conflicts between different sources.

If you're responsible for SEO / GEO, I'd suggest starting to expand your monitoring targets from "does the brand name appear" to "can AI accurately answer product questions." Page assets in traditional search are still important — it's just that now you also need to consider how these assets will be understood and reorganized once they enter the AI customer journey.

Twenty years ago, we studied how to help people find a page.

Today, we also need to start studying: when people aren't in a hurry to leave the AI interface, does the brand have good enough answers to make them stay?

Related Questions

What Are Sponsored Agents?

Sponsored Agents is a ChatGPT advertising experience OpenAI is currently testing. Users can enter a clearly labeled brand-sponsored Agent conversation from a relevant ad and continue asking about products or services; it's currently in limited alpha testing with select US advertisers only.

Do Sponsored Agent Responses Affect Regular ChatGPT Answers?

According to OpenAI's current description, the two are separate. The brand-sponsored conversation of a Sponsored Agent is distinguished from ChatGPT's independent answers, and is also separate from the user's existing ChatGPT conversation.

Will AI Agents Replace Landing Pages?

There isn't enough evidence to judge that at present. Sponsored Agents currently still allows consumers to go to the brand website at an appropriate stage; it's more appropriate to understand it as AI conversations starting to take on some explanation and comparison tasks, rather than Landing Pages disappearing.

What Is the Relationship Between GEO and Brand AI Agents?

GEO should focus not only on whether the brand appears, but also on whether AI can find, understand, and accurately express brand information. When consumers ask continuous questions in the AI environment, product facts, evidence, comparison information, and applicable boundaries all become important brand knowledge assets.

What Are "Conversational Brand Assets"?

It's not simply adding a chatbot, but organizing brand facts, product relationships, evidence, and boundaries to a degree sufficient to support continuous Q&A. When consumers rephrase the question, clear and consistent information still serves as the basis for answers.

Should Brands Start Doing Sponsored Agents Now?

Sponsored Agents is still in limited alpha, so at this stage it's more worthwhile for brands to first check whether their website, product pages, FAQ, and external information are complete and consistent, and to observe how the brand actually appears in AI environments like ChatGPT, Gemini, Google AI Mode, and Perplexity — then decide next steps based on the data.