ChatGPT enters the advertising world: this is how OpenAI intends to conquer brands (and what it means for cosmetics)

The way people search for information is changing, and ChatGPT has placed itself at the center of that shift. AI itself has become a search engine, the trusted search engine, even snatching this position from Google in some generations. More and more users turn to the chatbot to resolve doubts, compare products, plan trips, or decide on a purchase, and in that type of conversation, the user usually explains, in great detail, what they need. OpenAI has decided that there is also room for brands there and has begun to build its own advertising business.Germany, laboratory for ChatGPT's new advertisingOpenAI has started rolling out ads in 31 European markets, including Germany, after having launched this business avenue in other countries in early 2026. The move places ChatGPT within the map of digital advertising media and supports, albeit with different rules than those of what we might now consider more "traditional" media: the ad appears within a conversation in which the user has already explained, with considerable precision, what they are looking for.Dave Dugan, Vice President and Head of Global Advertising Solutions at OpenAI, who joined the company in March 2026 after more than a decade at Meta, has detailed in an interview with HORIZONT how this business is taking shape. According to him, Germany is a key market because it has the largest weekly user base of ChatGPT in Europe, with year-on-year growth exceeding 50%. Practically all Germans between 18 and 34 years old use the tool every week, and in the 35 to 44 age group, approximately two out of three people do so.With such a user base, OpenAI has detected that around 20% of conversations have a direct commercial intent: from choosing a holiday destination to comparing tech products or looking for a recipe. And unlike a traditional search, which usually reduces to two or three words, queries in ChatGPT average between 10 and 40 words, which leaves on the table a much greater amount of context, budget, conditions, and preferences than a conventional search engine.How ChatGPT decides which ad to showFor an ad to appear, OpenAI evaluates the relevance of the creative (image and text) to the user's query, in addition to the quality of the landing page. Added to this are the so-called "contextual clues": groups of five to ten words with which the advertiser explains what makes their product special.As for the buying model, it works through a second-price auction, with the option of working with CPM or CPC. CPC, launched just ten weeks after CPM, already accounts for most of the reserved volume, and both have been joined by oCPC (optimized cost per click), designed to direct ads towards users with a higher probability of conversion. In measurement, advertisers receive data on impressions, clicks, and conversions, and OpenAI has added a conversion API to more accurately track what happens after the click. Confidentiality of the conversation and "separate" responses from advertisingOpenAI insists on two principles. The first is "conversation confidentiality": the company assures that it does not share with advertisers the information that users type in the chat, and leaves it up to each person to deactivate memory or switch to Plus or Pro plans to avoid seeing ads.The second is the independence between the model's response and the ad. According to Dugan, there are two separate systems: one generates the organic response and the other manages the advertising, without influencing each other, something that is also visually reflected in the interface. For Dugan, the true value for brands lies in follow-up questions: when the user has already received a first recommendation and continues to refine what specific model, what features, what price is when the most relevant advertising opportunity appears.Agencies, formats, and the user sets the limitIn Germany, OpenAI already works with Publicis, WPP, Omnicom, Mediaplus, Havas, and Dentsu, and more than two dozen partners who can buy advertising via API. The company is also preparing a self-service platform for smaller advertisers. For now, in Europe the format is simple: image and text display, while in other markets there are already product catalogs for retailers and carousel ads. The idea, according to Dugan, is to advance little by little so as not to "overwhelm users."OpenAI monitors whether exposure to ads affects usage frequency or the likelihood of a user returning, and so far, according to the company, these metrics have not deteriorated in markets where advertising is already present. With over 1 billion weekly active users as a starting point, ChatGPT's advertising business is still under construction, but it already has clear pieces on the table: CPC, oCPC, conversion APIs, agency integrations, and new product formats.The context: AI is already the operating system of advertisingWhat OpenAI is doing is not an isolated case. Various industry analyses place artificial intelligence as a central piece, no longer just a simple support, of the advertising ecosystem by 2026. Trend reports point to a hybrid model where AI handles segmentation, optimization, and real-time analysis, while human teams continue to define strategy and brand identity.The deepest change, however, has to do with the purchase journey itself: increasingly, this no longer starts in a traditional search engine, but in a conversation with an AI assistant. Users are moving from "searching" to "asking," and this forces brands to structure their content to be useful not only for people but also for the AI systems mediating that conversation. In parallel, AI-generated creativity, real-time personalization, and co-creation with audiences appear among the trends that will mark marketing this year.Cosmetics: the sector where AI advertising already has real casesIf there is one industry that has an advantage in combining personalization, data, and artificial intelligence, it is ours, the cosmetics industry. Some examples illustrate where the sector is heading: L'Oréal uses AI to target what its CEO for Central America and the Andean Region describes as the beauty "consumer funnel": algorithms help decide what advertising to show, what product recommendation fits each person, and at what point in the purchase journey for makeup, skincare, or hair it is convenient to appear. The company feeds these models with data from its more than 40 brands and then adapts them to the identity of each one. It also developed ModiFace, an augmented reality application that allows users to virtually try on makeup before buying. Sephora combines AI and retail media: it uses its customers' data to personalize advertising campaigns with a more emotional than invasive narrative, in addition to anticipating trends and adjusting stock in real time. Neutrogena launched Skin360, an application that analyzes the state of the skin to recommend specific products and routines. Fragrance brands have developed devices connected to mobile apps where an AI system suggests personalized fragrance blends according to the moment and occasion. A pioneering chatbot from a makeup brand allowed users to virtually try products and complete the purchase without leaving the messaging app, anticipating years ago the conversational logic that ChatGPT now pursues.What all this means for a cosmetics brandIf ChatGPT starts showing advertising within conversations where someone asks, for example, "what moisturizer do you recommend for combination skin with a tendency to blemishes and a budget of 30 euros," the parallel with what L'Oréal or Sephora already do is direct. Some practical implications for the sector: Product content must be prepared to be "read" by an AI, not just by a search engine: clear product sheets, well-described ingredients, indications of what type of skin or need each product serves. The "contextual clues" proposed by OpenAI are an opportunity for a cosmetics brand to explain in a few words what differentiates it—active ingredients, certifications, target skin type—within an advertising auction. The moment of appearance matters more than the first result. In cosmetics, the user probably starts with a broad question ("best anti-aging serums") and refines it to a very specific need; there, in the second or third question, is where a brand can have more options to appear with real relevance. Personalization is no longer a plus, it is the consumer's expectation. Various industry studies indicate that a large majority of consumers expect personalized shopping experiences, so conversational advertising naturally fits with an industry accustomed to skin diagnoses, tone tests, or tailored recommendations. Trust is the most fragile asset. In a sector where AI already analyzes faces, skin tones, and behaviors, transparency about the use of that data and about what is organic content and what is advertising will be as important as the campaign's effectiveness itself.ChatGPT arrives somewhat late regarding cosmetics in the use of AI to personalize the consumer experience, but it can become another channel to apply that same logic: understand the conversation, appear at the right moment, and let the user decide how much advertising they are willing to accept in exchange for more precise recommendations.

ChatGPT enters the advertising world: this is how OpenAI intends to conquer brands (and what it means for cosmetics)

With more than 1 billion weekly active users as a starting point, ChatGPT's advertising business is still under construction, but it already has clear pieces on the table