What is AI Search?
How am I performing in AI Search? It’s the question coming up the most in client conversations. To answer the question it’s important to define what AI Search is. AI Search is leveraging an AI model to perform a search query but instead of keywords (SEO) you provide a prompt (command or question) for the AI to execute. The AI will sift through data it has been trained on including websites and social media, especially Reddit. It will also sometimes perform a search query to help it have access to the latest information. Naturally, Google’s Gemini does this automatically and there are AI Overviews for Google Search which pull answers from several different sources and composes something suitable to address the question/command.
A December 2025 study via Search Engine Land shows that Google increased AI Overviews before pulling back. Specifically, in January 2025 6.5% of queries triggered an AI Overview, increasing to just under 25% of queries in July before falling back to a bit less than 16% of queries in November 2025. Also an interesting thing to note from this study is that the overviews started almost purely for informational keywords (94% in January 2025) to more mixed (57% informational in October 2025, 18% commercial and 14% transactional). What this means – Google is doing a lot of experimentation which means a volatile environment but lots of opportunity as well.
According to SEO & AI Search leader, Lily Ray, marketers and business owners should avoid some quick-win tactics like “best of” listicles that go against Google’s underlying EEAT (Experience-Expertise-Authoritativeness-Trustworthiness) principles for quality content. In fact, Google is increasingly adding more reference links to AI overviews so that 1) searchers can see where the data for the AI overview came from and 2) searchers can view the content directly on the provider’s page as they would traditionally do for a click in a web search. The bottom line for Google – the desire for featuring high quality content is still at the base for how Google operates and maximizing your SEO is still important for Google search traffic as well as other AI models/agents that rely on Google’s data.
AI Search is not uniform. Each model has a different flavor for how it collects and collates data as well as how it arrives to an answer (algorithm). According to Profound, ChatGPT’s top sources are Wikipedia (5% of all citations), Reddit (3%), and Reuters (1%). According to Claude’s Transparency Hub, “Claude Sonnet 4.6 was trained on a proprietary mix of publicly available information on the Internet as of May 2025” – which means a lot of times it is performing a search query to get the latest data (after May 2025). You absolutely want the most up-to-date data for accuracy and this also goes to show how strong existing search databases, especially Google, are. And, therefore, SEO is still the primary weight for GEO.
How Do I Maximize My Visibility in AI Search?
The answer is largely to do SEO well. Like traditional search models, AI wants to provide or help direct their users to the best answer. That means the answer must come from somewhere qualified or of quality. Quality content is largely original content that contributes. Informational keywords are a great example of low hanging fruit – how can you help answer the question in a way that helps the user most effectively? According to The DO Lectures, the 7 principles of writing to create outstanding content every time are: clarity, credibility, originality, authenticity, productivity, brevity, and creativity. How good is the quality of the content you’re providing? Would it be something you’d find value in if you didn’t know anything about your business? Are you contributing to the quality of the internet? Some gut check questions to ask before coming up with a content strategy or hitting publish on a particular piece.
The other factors relate to the relative strength/perception of your brand. Are people tagging your company in their Instagram or Facebook posts? Are people writing love letters to your brand on Reddit and other reviews across the internet? Are people generally speaking positively or negatively about you on the internet? This work has been historically called social monitoring and was tied to traditional public relations work. Now, with how fast AI evolves, it falls under brand, specifically brand perception. There are different tools like Semrush or Profound who will provide you with an AI visibility score as well as cited sources but what is your brand? Is what you think of your brand the same as your customers see it? Your employees? AI will find out and factor the perception points into its answers.
What to do if brand perception is low or not at the level you believe it deserves to be or should be? Like SEO, brand perception is a slow/cumulative work. When speaking with clients I make reference to a steady gym strategy. Are you exercising consistently and aligning your workouts/diet with what your body needs? Things are probably going well for you. Same thing for SEO and brand perception. Are you creating quality content tied to what your market is looking for? Are you a brand that’s accessible and engaging? Are you providing a quality product or service? Do your reviews reflect the quality of the service or product you offer? Digitally it’s quite the game of whack-a-mole that could quickly become exhausting. That being said, brand equity, or the strength of your brand, is typically determined by the underlying principles of what you stand for. Tools can help you pinpoint and address bad perceptions as they crop up but a brand focused on quality and who engages their market for brand/social proof will typically get to the desired point easier and faster.
