Conversation-Based Advertising Formats
Conversation-based advertising formats are ads that appear within a conversation with a computer program, for instance in a chat window. Instead of a rigid advertising banner, the advertising becomes part of the dialogue and is tailored to what the user has just asked.
More and more people no longer search for information via a list of links, but instead ask a program a question in natural language and receive a written answer. Conversation-based advertising formats are ads that show up precisely within this conversation. If someone asks for a good travel camera, the answer may contain a paid product suggestion. Sometimes it’s a brief note beneath the answer, sometimes a recommendation woven into the middle of the text, sometimes a follow-up question like: “Would you like to see offers?” The difference from a classic ad lies in its form: here, advertising doesn’t appear as a colorful rectangle at the edge of the screen, but as a contribution to the ongoing dialogue.
What is at stake for search engines and the advertising market
The largest share of money on the internet comes from search ads. Anyone who searches Google for “buy running shoes” sees paid results at the top. This business only works because people actually look at search results at all. If a chat program answers the question directly, the results list disappears - and with it, the ad space.
This is exactly why the major providers are working on new formats. For them, it’s the decisive question of how expensive language models can be financed at all. Such models are programs that have learned to continue language from vast amounts of text. Every single answer costs computing time, and therefore money. Only a few users pay subscription fees, whereas advertising reaches everyone.
For companies selling products, this shifts the competition. Until now, it was about ranking high in the results list. In the future, it will be about being mentioned in the answer at all. Whoever is missing there practically doesn’t exist for the user, since an answer often names only two or three options.
How an ad makes its way into the dialogue
Technically, this usually happens in two steps. First, the system evaluates what the conversation is currently about and detects a purchase intent. Then an auction takes place: advertisers have previously set how much they are willing to pay for certain topics. The winner gets to have their product woven into the answer.
The text of the ad itself is often only generated in the moment. The model receives the advertiser’s product data and formulates a sentence from it that fits the question. That’s why the same ad can look different for two different users. Think of it like a salesperson who has a fixed range of products but explains their recommendation differently to each customer.
This gives rise to the core problem of these formats: the line between advice and advertisement blurs. With a banner, everyone knows it’s an ad. A recommending sentence in the middle of an answer, by contrast, comes across as neutral information. Regulators therefore demand clear labeling, for instance with the word “Ad” placed directly at that spot. Incidentally, a common misconception is that the ad retrains the model’s content. In fact, it is merely injected additionally - the underlying model itself remains unchanged.
Where these ads already appear today
The development is most visible in search engines with AI-generated answers. Google displays summarized answers above the classic results and mixes in paid hints there. Chat providers such as Perplexity have also launched advertising programs. At OpenAI, the company behind ChatGPT, advertising was excluded for a long time, but it is now being tested.
Besides this, the format also appears in messenger services and customer support. An airline’s chatbot answers a question about baggage and then suggests travel insurance. Such sales suggestions within a conversation existed even before today’s language models - they were just written according to fixed rules.
In business news, you’ll usually encounter the term in connection with revenue expectations. Analysts estimate how much advertising money is shifting from classic search to chat interfaces. This concerns a market on the order of hundreds of billions of dollars per year. So when you read about “Conversational Ads,” it refers less to a technology than to the question of who will collect the internet’s advertising revenue in the future.
