Search behavior has quietly shifted, and most business owners haven’t caught up yet.
A growing share of people no longer type a query into Google and click through a list of blue links. They ask ChatGPT, Perplexity, or Gemini a question directly “who’s the best insurance broker near me,” “which kitchen renovation company should I hire,” “how do I find an immigration consultant” and take whatever answer comes back. No scrolling. No comparing ten tabs. Often, no visit to a company’s website at all.
For a business, that’s a problem if the AI never mentions you.
A recent look at roughly 200 small and mid-sized businesses spanning home renovation, insurance, healthcare, local services, and immigration consulting offers a rare, concrete look at what separates companies that show up in AI-generated answers from those that have effectively become invisible to them, even while still ranking fine on Google.
The gap between “ranking well” and “being mentioned”
Traditional SEO was built around one goal: rank near the top of a Google results page. That still matters, but it’s no longer the whole picture. Large language models don’t browse a ranked list of pages and pick the top one they draw on whatever content has been structured clearly enough, and cited widely enough, for the model to treat it as a trustworthy answer.
That’s a different bar. A business can rank on page one of Google for its core keywords and still never get named when someone asks an AI assistant the equivalent question conversationally.
Case in point: a kitchen and closet renovation company in the sample published 58 pieces of structured content and built out 187 internal links across its site not to chase a single keyword, but to give an AI model enough context to understand and describe what the business actually does, in enough different ways, that it could be surfaced across a range of related questions.
A convention centre in the group focused less on content volume and more on local signals its Google Business Profile activity alone generated an estimated 8,700 local actions, the kind of real-world engagement data that increasingly feeds into how AI models judge which local businesses are legitimate and worth recommending.
An immigration consultancy and a perfume retailer in the sample were each tracked as being surfaced across roughly 100-130 distinct AI-generated queries, a rough proxy for how often a model might mention them when someone asks a related question, even if that person never searches the brand name directly.
Why this is happening now
The shift has a straightforward explanation. AI assistants are trained and continuously updated on the open web, and increasingly on real-time retrieval from it. If a business’s website doesn’t clearly explain, in plain structured language, what it does, who it serves, and why it’s credible with the kind of detail a human expert would actually write, not thin, keyword-stuffed filler there’s simply less for a model to work with when someone asks a related question.
That also happens to line up with what Google itself has been pushing search engine optimization toward for the past two years, through its “helpful content” guidance: reward pages written by people with real experience and expertise, for actual readers, rather than pages engineered purely to game rankings. The businesses that adapted early for one didn’t have to start from scratch for the other clear, genuinely useful, well-organized content tends to perform for both a search engine’s ranking algorithm and a language model’s citation behavior, because both are ultimately trying to reward the same thing: content that actually answers the question well.
What the businesses that adapted have in common
A few patterns show up repeatedly across the companies that managed to increase their visibility in AI answers:
Consistency mattered more than any single piece of content. Businesses that published a steady cadence of new, specific material service pages, local guides, FAQ-style content answering real customer questions saw broader coverage across different AI queries than those that published once and stopped.
Local and structural signals counted as much as written content. Google Business Profile activity, internal linking, and basic technical health (fast-loading pages, clean site structure) showed up as consistently as blog content in the businesses that improved.
Tracking actually mattered. Several businesses in the group used automated tools to monitor whether and how they were being mentioned across ChatGPT, Perplexity, Gemini, and Google’s AI-generated overviews treating AI visibility as something to measure, not something to hope for.
Among the tools used by businesses in this sample was UpliftAI SEO Software, a platform built specifically to handle this newer version of the problem generating structured content, publishing it across a business’s website and Google Business Profile, and tracking whether that content actually gets surfaced when AI models are asked relevant questions. A rundown of similar tools and how they compare is available in this breakdown of the best SEO automation tools currently built for this shift.
The takeaway
Nobody knows exactly how AI models will decide what to cite five years from now the systems themselves are still changing quickly, and no one, including the companies building them, claims to have fully figured it out. But the businesses in this sample that treated AI visibility as a distinct, measurable problem rather than assuming that ranking well on Google would automatically carry over appear to be the ones adapting fastest to a search landscape that no longer starts and ends with a results page.
For small businesses in particular, that’s arguably an opportunity as much as a threat. Being genuinely well-documented, specific, and locally credible may matter more to an AI model than having the biggest advertising budget which, at least for now, levels a playing field that used to be tilted toward whoever could outspend everyone else on ads.
Disclaimer: The information provided in this article is for general informational and educational purposes only and should not be considered professional, financial, legal, medical, business, or other specialized advice. While reasonable efforts have been made to present accurate and up-to-date information, facts, statistics, opinions, and recommendations may change over time and may not apply to every individual or situation. Examples and results mentioned are provided for illustrative purposes and should not be interpreted as guaranteed outcomes. Readers should independently verify important information and consult an appropriately qualified professional before making decisions based on the content. The publisher assumes no responsibility for losses arising from reliance on this information.