OpenAI is beginning to describe ChatGPT advertising with a comparison that should get the attention of both search marketers and social advertisers.
OpenAI CFO Sarah Friar has framed the long-term opportunity as something resembling a combination of Google and Meta: Google’s ability to capture explicit, high-intent demand, Meta’s ability to understand personal context, and an additional layer that neither comparison fully captures — ChatGPT’s conversational memory.
Speaking at the Goldman Sachs Communacopia + Technology Conference, Friar described the advertising opportunity around the combination of search-like intent and richer user context, according to Business Insider. It is a framing she has used before. In a June interview on the All-In podcast, Friar said that if Google and Meta “had a baby,” the result would look like ChatGPT, pointing to high intent, demographic context and memory as potentially powerful ingredients for advertising.
The comparison is more important than the metaphor.
Google built one of the world’s most valuable advertising businesses by monetizing what people say they want now. Meta built another by predicting what people may want based on who they are and how they behave. ChatGPT potentially sits in a third position: users can explicitly describe what they are trying to do while the system may also, when personalization controls permit it, have relevant context from earlier interactions.
That could make conversational advertising neither conventional search nor conventional social advertising.
Google’s advertising advantage begins with declared intent
The enduring strength of search advertising is that the user does much of the targeting work voluntarily.
Someone searching for “best CRM for a 20-person sales team” has declared a commercial problem. A query for “flights from Milan to Tokyo next month” communicates destination, origin and approximate timing. “Emergency plumber near me” can signal both category and urgency.
Advertisers value those moments because the need is explicit.
Google has spent decades building an auction around that declared demand. Modern Google Ads uses far more than keywords, but the underlying advantage of Search remains straightforward: the user tells the platform what they are looking for at the moment the ad opportunity appears.
Friar’s comparison suggests OpenAI sees a similar asset inside ChatGPT.
Users are not merely scrolling through a feed. They are frequently asking for help with a problem, researching an option, comparing alternatives or planning an action.
A conversation can reveal more than a search query
The difference is that a ChatGPT request does not need to compress intent into a handful of words.
A search user might type “best laptop for video editing.” A ChatGPT user can explain that they edit 4K footage in DaVinci Resolve, travel frequently, care more about battery life than gaming performance, have a $2,000 budget and already own a particular monitor.
Both interactions can express commercial intent. The conversation can make the constraints surrounding that intent much more explicit.
That distinction is critical.
It would be simplistic to say conversational intent is always stronger than search intent. A user typing an exact product model plus “buy” into Google may be extremely close to a transaction. The advantage of a conversation is not automatically higher intent; it can be higher context.
For an advertising system, knowing what someone wants and why they want it can create a much richer relevance problem than matching an ad to a short query.
Meta’s advantage is context without requiring a search
Meta approaches advertising from the opposite direction.
A Facebook or Instagram user does not need to explicitly search for running shoes, a holiday or business software before an advertiser can reach them. Meta’s systems use signals and machine learning to predict which ads may be relevant while users browse social products.
That model gives advertisers enormous discovery power. It can create or capture demand before the user has translated it into a search query.
Friar’s analogy is that ChatGPT can potentially combine part of that contextual understanding with the explicit intent expressed inside a conversation.
The result would not literally be Google targeting plus Meta targeting copied into one interface. OpenAI’s current advertising product has its own privacy rules, controls and targeting mechanics. But strategically, the combination is clear: a user can be actively expressing a need while the system also has contextual information that can make that need easier to interpret.
Memory is the element that makes the comparison more interesting
ChatGPT’s memory changes the advertising discussion because context does not necessarily end when one prompt ends.
OpenAI’s current Ads in ChatGPT documentation says that when ad personalization is enabled, the system may use select signals from a user’s broader ChatGPT experience to make ads more relevant. If memory is enabled, OpenAI says saved memories and recent chats may be referenced when selecting an ad.
That creates a potential form of continuity that a single search query does not contain.
A user might discuss plans for a marathon over several conversations, mention a preference for low-impact training and later ask for advice about replacing running shoes. The current question supplies immediate intent; previous context can potentially help the system understand the situation more precisely.
That is the layer Friar is pointing toward when she highlights memory alongside search-like intent and personal context.
Advertisers do not get access to the user’s memories or conversations
This distinction is essential because “using memory for ad relevance” can easily be misread as “giving advertisers access to memory.”
OpenAI says it does not do that.
The company’s advertising principles state that conversations remain private from advertisers and that OpenAI does not sell user data to advertisers. Advertisers do not receive a person’s chat history, memories or private conversation content.
The relevance decision happens on the platform side.
That means an advertiser may benefit from an ad being matched to a contextually relevant conversation without receiving the underlying private context that made the match possible.
This is a fundamental boundary in understanding ChatGPT Ads. The system may know more about the conversational situation than the advertiser does.
Personalization is also a user-controlled layer
The Google-plus-Meta analogy should not imply that every ChatGPT ad automatically uses a persistent personal profile.
OpenAI says users can control ad personalization. Broader ChatGPT signals and memory can be used for ad selection when the relevant personalization and memory settings allow it. Users can turn personalization off, and Temporary Chats do not show ads.
Those controls mean the available context can differ between users and sessions.
The immediate conversation can still communicate substantial commercial intent even without persistent personalization. A detailed live discussion about choosing a mortgage provider, planning a holiday or buying enterprise software can contain enough context to make a relevant ad possible without drawing on older conversations.
Memory adds another layer; it is not the only layer.
The advertiser’s unit of targeting may shift from keyword to situation
This is where the model becomes strategically different for marketers.
Search advertising trained generations of advertisers to think in keywords and queries. Social advertising trained them to think in audiences, creative and behavioral prediction.
Conversational advertising encourages a different unit: the situation.
An advertiser may care less about whether a user typed the exact phrase “best accounting software” and more about whether the conversation describes a five-person company struggling with invoices, international payments and tax reporting.
The relevant commercial context is not a string. It is a problem state.
OpenAI’s advertising system has been developing around this idea. In its May expansion of ChatGPT Ads, the company added self-serve buying, CPC bidding and broader measurement while emphasizing that conversations and personal details are not shared with advertisers.
The platform is therefore building conventional advertising infrastructure around a very unconventional source of intent.
ChatGPT could see intent before it becomes a conventional search
Another difference is timing.
Search advertising often captures demand after the user has reduced a problem to a query. Conversational AI can participate earlier in the reasoning process.
A person might begin by asking whether they should repair or replace a washing machine. Several turns later, the discussion may shift toward brands, dimensions, energy efficiency and budget. Only then does a purchase become the likely next step.
ChatGPT can observe that decision being formed inside the conversation.
For advertisers, that potentially creates opportunities between broad awareness and final search intent — a stage where the user is actively reasoning rather than passively consuming content.
That does not eliminate Google’s advantage at the transaction-ready end of Search or Meta’s scale in discovery. It creates another commercial moment.
Context can improve relevance, but it also raises the standard for trust
The same characteristic that makes conversational advertising potentially valuable makes it unusually sensitive.
People tell an AI assistant things they would never put into a search box or public social post. They can discuss budgets, family plans, career decisions, anxieties and highly personal constraints while asking for help.
OpenAI therefore has a narrower margin for error than a conventional advertising platform.
The company says ads are not eligible to appear near sensitive or regulated conversational topics such as personal health, mental health or politics, and it maintains separate rules for advertiser eligibility in regulated categories.
The larger strategic challenge is preserving the feeling that ChatGPT is helping the user rather than steering the conversation toward whoever paid for placement.
OpenAI says ads cannot influence ChatGPT’s answer
That boundary is central to the current product design.
OpenAI says sponsored placements remain separate from the answer and advertisers cannot pay to change what ChatGPT says. Ads are clearly labeled, and the advertising system is not supposed to determine the organic response.
This separates the monetization layer from the recommendation layer in a way that is particularly important for conversational AI.
If users began to suspect that product recommendations inside an answer were being rewritten to favor advertisers, the value of the assistant’s advice could deteriorate quickly.
OpenAI’s advertising opportunity therefore depends on monetizing context without allowing the advertising auction to purchase the model’s judgment.
Google and Meta are analogies, not a finished product specification
Friar’s comparison should also be read as a strategic vision rather than a claim that ChatGPT Ads already has every capability of Google Ads and Meta Ads.
Both incumbents have spent decades building auction systems, conversion infrastructure, advertiser tools, fraud defenses, attribution systems and optimization models at enormous scale.
ChatGPT’s advertising platform is much younger.
OpenAI began with limited sponsored placements and has been progressively adding buying and measurement capabilities. Its self-serve Ads Manager and CPC bidding only arrived in 2026.
The interesting question is therefore not whether ChatGPT has already matched Google and Meta operationally. It is whether the underlying conversational data model gives OpenAI a distinctive foundation on which to build.
The strongest signal may be intent plus constraints
For marketers, one of the most valuable aspects of conversation is that users often volunteer the constraints that determine whether a product is suitable.
Price ceiling, location, compatibility, experience level, urgency, preferred features and previous purchases can all emerge naturally while the user works through a decision.
A search query may express some of these constraints. A long conversation can express many simultaneously.
That potentially allows advertising relevance to move from “this person is interested in running shoes” toward “this person needs a stable daily trainer under a specific budget for a particular training pattern.”
The second description is commercially much richer.
It also means advertisers may need to improve landing pages and product information. A system capable of understanding detailed intent will have little reason to reward an ad whose destination cannot satisfy the detailed need.
The new competition may be over who understands the customer’s decision best
Google’s advertising empire was built around organizing demand expressed through queries. Meta’s was built around predicting demand from identity, behavior and content consumption.
ChatGPT introduces the possibility of understanding demand through dialogue.
That dialogue can reveal immediate intent, changing preferences and the reasoning behind a purchase. With user-controlled personalization, some relevant context can also persist beyond the current exchange.
This does not guarantee better advertising. More context can still produce poor predictions, irrelevant placements or uncomfortable personalization. The platform must also maintain strict privacy boundaries and avoid turning sensitive conversations into commercial opportunities.
But the advertising asset is structurally different.
ChatGPT Ads may be most powerful when they are neither Google nor Meta
Friar’s Google-and-Meta comparison is useful because it explains the two established strengths OpenAI hopes to combine: declared intent and personal context.
The more interesting possibility is what happens when conversational memory and multi-turn reasoning are added to that equation.
A search engine usually sees a query. A social network builds a model of interests and behavior. A conversational assistant can potentially see the problem being explained, refined and solved in real time.
That could create an advertising system organized around the decision itself rather than simply the keyword that expresses it or the audience predicted to make it.
OpenAI still has to prove that this produces better results for advertisers without damaging user trust. Its current policies make clear that advertisers do not get private conversations or memories, personalization remains controllable, and sponsored placements remain separate from ChatGPT’s answers.
If those boundaries hold, ChatGPT Ads could evolve into something more specific than “Google plus Meta.” It could become an ad platform built around a signal neither incumbent originally designed for: a persistent conversation about what the customer is actually trying to accomplish.