SearchGPT, OpenAI’s new search engine prototype, has created a wave of enthusiasm. It has also sparked several pressing questions: how will AI reshape the search landscape, and what does it mean for search engine optimization (SEO)?

For years, SEO has been an art largely dedicated to optimizing content for Google. Now, with generative engine optimization (GEO), the rules may be changing.

Google is still a dominant force in search, commanding 89% of the market worldwide in 2023-2024, according to StatCounter. In May, the Mountain View giant introduced AI Overviews to US users and has since expanded the feature to over 100 countries. Meanwhile, publishers and other companies are beginning to notice new audiences coming from AI search tools like Perplexity, as well as chat assistants powered by large language models (LLMs), such as ChatGPT, Microsoft Copilot and Meta AI.

For IBM’s own digital platforms, this AI-driven audience is still small but undeniably valuable. Matthew Blackshaw-Marschinke, Manager of Global SEO at IBM, estimates that the AI-driven audience is 100 times smaller than the traffic Google delivers—though more deeply engaged. “When comparing the two [traffic sources] with the actions we care about—downloading white papers, completing web forms or signing up for newsletters—the conversion rate for AI platforms is double that of Google search,” says Blackshaw-Marschinke. “It’s still early days for the technology, but these users show higher intent and growing trust in the content.”

For the first time in 20 years, generative AI is opening up new opportunities in the search market. Perplexity, an AI-powered search engine launched in December 2022, says it now handles over 100 million queries a week. Another AI search engine, You.com, which debuted in 2020, added multimodal chat search in 2023, allowing users to get outputs beyond text, such as images and charts.

Among the latest entrants is BeaGo, introduced this fall by Rhymes AI, a startup co-founded by Anita Huang, Junnan Li and venture-built by Kai-Fu Lee. Focused on LLM multimodality, the company created Aria, an open-source, multimodal native mixture-of-experts (MoE) open model, and Allegro, a text-to-video model.

“It's very competitive now with AI search in the space, but I still feel there are different segments of users and different use cases that may allow more players to satisfy consumers’ needs,” says Huang, a former executive at Yahoo! Taiwan and Google China.

BeaGo has a simple interface. Its goal? Providing users with a single, high-quality answer.

“We wanted to be able to provide something—like a smart beagle dog that fetches you something really quick—so you have a condensed and structured overview of what you are searching for on the go,” Huang says. Huang also highlights how BeaGo enhances its answers with supporting images and graphs—with plans to add short videos as well. “It marries the intelligence of large language models with the power of web search,” she says. “We collect data on the fly and use LLMs to synthesize and present the information in the best way possible.”

Simplicity in content and design is also key, says Chris Andrew, co-founder and CEO of Scrunch AI. His startup helps Fortune 500 companies navigate the new digital landscape dominated by generative AI.

“My mentality is that consumers always prefer simplicity. They'd rather get an answer, than 10 links,” he says. “AI wants simple, accurate, structured content. Interactive or complex layouts make it harder for AI to extract answers. Clean, well-organized text ensures that your content gets surfaced.”

Of course, if users get a satisfactory answer from Google AI Overviews or SearchGPT, they might not even click on a link. And though Google recently added clickable links on the right side of AI Overviews, some still fear a loss of traffic to their pages.

“I have been using AI Overviews, and I absolutely don't click as much," observes Lily Ray, Vice President of SEO Strategy and Research at Amsive, a marketing agency.

As Ray sees it, the shift to AI search engines will push brands toward a more comprehensive marketing approach. “You have to manage your brand’s reputation, what people are saying in their reviews, your social media presence,” she says. “All of these holistic marketing strategies affect how you’re represented in large language models.”

Blackshaw-Marschinke suspects that chatbots will favor high-quality content. "The first key idea, in my opinion, is 'information gain'—the concept that, as a publisher, you need to create content with unique characteristics that go beyond the general consensus already present in the model," he says.

Andrew stresses that the search revolution is only just beginning. The rise of AI agents, which can access content and take action on a user’s behalf, could force businesses to rethink their GEO once again. The brand and the audience won’t have a direct conversation, and it will be harder to influence the agents.

“These models will keep getting better,” he says. The challenge is whether brands and marketers can keep up.