Marketers Prove Skeptical of Advertising In Chat-Based Search Engines

As Microsoft and Google race to integrate the next wave of language artificial intelligence into their respective search engines, the push has left agencies with questions about what these new conversational platforms will mean for how ads are served. So-called large language model AI systems like ChatGPT could offer users a more interactive way to access information on the web. But how Microsoft and Google will seek to monetize their huge investments in this technology with paid placements is so far unclear to ad buyers. Microsoft seems to be considering advertisements in the form of links listed at the end of the answer to a given query, according to what advertisers have seen within the current Bing AI platform. Meanwhile, buyers say Google has not disclosed any details about how ads will work within its own Bard AI, a chatbot designed to compete with Microsoft’s new GPT-powered Bing. The Rise of AI Content Generation Stirs Brand Reputation Fears But concerns remain about how monetization will fit into the flow of chatbot conversation, if it will prove effective and whether the technology might fall short with people in the same ways that previous attempts to reinvent the search experience have. “I’m most skeptical about the potential of ChatGPT to disrupt ads in a meaningful way,” said Sam Tomlinson, executive vice president at ad agency Warschawski. Positions of ad formats and new data The ad format that Microsoft has already suggested placing an ad at the end of a chatbot query, leaves marketers unconvinced. It’s not clear at what point in the conversation an ad might be served, said Aaron Levy, vp of paid search at performance agency Tinuiti. The pricing is also unclear, given that there would likely be fewer impressions for any given chat compared to a search query since conversational formats are more complex. Marketers can also imagine other ways AI chat-driven search could lead to advertising. Bing’s chatbot could include links that go to relevant Bing searches, where users would then be served with ads, said Michael Cohen, evp of performance media at Horizon Media. The technique, known as search arbitrage, has also been employed by Ask.com and Yellow Pages. But adopting ad formats similar to what's currently available is a missed opportunity. “If you are trying to disrupt search, you have to do it in a meaningful way versus a slapdash way like putting links at the bottom,” Tomlinson said. “[If you do that], you’ve just swapped out content at the top.” Advertisers also ponder what data Microsoft and Google will be able to access to power their tech. With chatbots, people will spend more time interacting with content on the search engine’s site versus mostly interacting with content once they click on a link, potentially providing the tech firms, and in turn, advertisers, new reams of data. “Can you show me the query?" Tomlinson said. “We could deploy that against other aspects of our strategy, like content strategy and search engine optimization and product strategy.” Comparisons to voice-based search For all the potential AI has to upend search, the format has proved very sticky over the years. Agency veterans tend to compare AI-based search to voice-based search. Tech giants similarly heralded the advent of digital assistants as a means to put a more conversational interface on a search into which ads could be interspersed. But the format never reached the type of consumer adoption it would have needed to become a major advertising channel in its own right. Generative AI-based search might not be entirely analogous, but it does share some of the challenges around placing ads in a more inherently conversational setting. “It's being marketed as a paradigm shift, and it kind of is, if it takes,” Levy said. “Voice search never quite got there. This isn't quite the same. It's a stepping stone instead of a big leap, like going screenless would have been with voice search.” One of the problems with voice search was that it lacked the same feedback loop by which Google text search learns from each user’s behavior and becomes more efficient as a result, according to Tomlinson. “By the time you got the answer, it was stupid,” Tomlinson said. “And you went to a regular Google search to get the answer.” With AI-based Custom Algorithms, Marketers Wring More Value From the Open Web Google's dilemma Google—which is a lot more reliant on text-based search for ad revenue—has also rolled out a variety of features in the past decade that have transformed the search experience to serve information more directly to people within its walled garden. Results formats like Featured Snippets, the Knowledge Graph sidebar and the People Also Ask box pull and highlight information directly from websites on the results page itself. Notably, none of these features offer paid advertising because they tend to be more strictly informational in nature, said Levy. Still, Google has been pushing further into ads that live exclusively on the results page in recent years, said Levy, including formats like Local Service ads, in which the entire transaction with a nearby small business takes place within Google’s site. Such offerings could provide a model for a more native form of conversational AI ad.