Is AI Visibility Just SEO With a New Name?

Partly, yes. Most of what makes a business visible in search also matters to AI systems, and Google treats optimizing for its AI features as SEO. The open question is narrower. When an AI assistant compares several businesses against one buyer’s requirements, does strong search performance tell you how that comparison will go? On its own, it doesn’t.

Google’s Position Is Reasonable

Google’s guide to optimizing for generative AI features on Google Search, last updated in July 2026, says SEO best practices remain relevant and that optimizing for generative AI search “is optimizing for the search experience, and thus still SEO.” It also tells site owners they can ignore content “chunking,” AI text files such as llms.txt, and the pursuit of inauthentic mentions.

That follows from how the product is built. AI Overviews and AI Mode run on systems Google spent decades refining to discover, index, rank and retrieve pages. A question can be broken into several related searches, which Google calls query fan-out, and the answer is assembled from what comes back. Google’s AI features depend heavily on the information its search systems can discover, index and retrieve.

The guide’s advice on content points the same way. It asks for “non-commodity” content, and notes that because its AI systems look at a variety of sources, “it can be helpful to have a unique viewpoint that stands out.” In effect, Google is telling businesses they don’t need a separate set of AI tricks. They need useful, original information, made available through sound SEO.

We agree with that. Two qualifications still apply. The guide is about Google Search, and Google’s dominance doesn’t settle the wider question. And “still SEO” describes the work. It doesn’t mean SEO’s usual measures will show how a business fares when an assistant compares it with its competitors.

ChatGPT and Claude Took a Different Route

ChatGPT and Claude began as conversational models and added web search later, the reverse of Google’s path. The difference is easy to overstate. Both assistants now search the web, and OpenAI says ChatGPT search “sometimes partners with other search providers,” rewriting a question into targeted queries and ranking results “using multiple factors.” Neither company publishes those factors, and there’s no reason to assume their retrieval works like Google’s.

The sources they draw on do differ. In an Ahrefs study of 15,000 prompts from August 2025, only 12% of the links cited by ChatGPT, Gemini and Copilot appeared in Google’s top 10 for the same prompt. The study compares the sources those assistants visibly cited with Google’s rankings for the original query, so it can’t show everything they retrieved along the way, and it’s a vendor’s snapshot of fast-changing products. Even so, it suggests that ranking well on Google and being cited by another assistant aren’t the same thing.

What differs more clearly is the interaction. A buyer can describe a situation in detail, add constraints as the conversation develops and ask for a comparison. The answer can arrive without a visit to any of the businesses involved. As the systems converge, that shape of interaction matters more than which company started where.

Being Encountered and Appearing Suitable

Picture two established contract manufacturers serving medical device companies. Both have credible websites and long track records.

The first is far more prominent in search. It ranks for the broad terms and is widely written about. The second is smaller, but its site documents exactly what it does: ISO 13485 certification, low-volume production runs, the materials it works with, and project write-ups that show each of those in practice.

A buyer asks an assistant:

We need a contract manufacturer for about 500 sterilizable device housings. ISO 13485 certified, experienced with low volumes, ideally in the Midwest. Who should we talk to?

The specialist could fare in three quite different ways, and none of them depends on its quality alone.

Its evidence might never be retrieved. If the assistant’s searches reach the prominent firm’s pages and not the specialist’s, the specialist isn’t in the comparison, however well documented it is. What the question asks for shapes what gets retrieved, but that doesn’t guarantee the best evidence is found.

Its evidence might be retrieved and still be outweighed. A firm that is described often across the web arrives with more context around it, and research on conversational search engines has found that models vary in how much weight they give product names relative to the content of the pages in front of them. Getting into the comparison isn’t the same as winning it.

Or the assistant might recognize the specialist as the closer match, because the buyer’s requirements are specific and the specialist has published checkable evidence that it meets them.

Same business, same website, three outcomes. Strong search performance doesn’t guarantee the third, and neither does publishing the right evidence. Even the third is only a starting point. The buyer would still need to verify certifications, capacity and availability directly. An AI recommendation isn’t supplier due diligence.

So there are two related questions here. What makes a business likely to be encountered? And what makes it appear suitable for a particular buyer once its evidence is considered? They draw on overlapping evidence. Specific requirements can steer what an assistant searches for, so evidence of suitability can affect retrieval. Prominence can shape the final answer, not only who gets found. Search performance answers the first question well. It says much less about the second.

Where the Feedback Loop Changes

This is the part I find most interesting, and the part where the evidence is thinnest.

Google’s search systems have long learned from how people interact with results. Google says it uses “aggregated and anonymized interaction data to assess whether search results are relevant to queries,” and the court in the US government’s antitrust case against Google described a ranking signal trained on months of user click data. On a results page, people choose among visible alternatives, and those choices, in aggregate, help shape what appears first.

In a conversational answer, some of that comparison happens before the user sees the alternatives. The assistant weighs the candidates and presents one or a few. The customer accepts the suggestion, rejects it or narrows the brief, often without visiting any of the businesses.

That doesn’t mean there’s no feedback. Continuing a conversation, accepting a recommendation or refining a requirement could all be useful signals to a provider, and the major providers say user feedback is used to improve their models. What isn’t publicly established is whether, or how, those signals affect which businesses are recommended next time.

What has changed is the behavior being observed and the point at which alternatives are shown or set aside. Whether that shifts the relative influence of search prominence and evidence of suitability is an open question. Retrieval still runs largely on search systems, where prominence plausibly matters a great deal. In one study of 10,000 answers generated through the GPT-4o API, page-one Google rankings correlated strongly with brand mentions, although a correlation doesn’t establish cause, and a mention isn’t a recommendation. Any shift may only matter for businesses that are encountered in the first place.

I wouldn’t put it more strongly than that. But it’s a question worth asking before assuming that the advantages built through traditional search will carry over unchanged.

A Good SEO Program Already Covers Much of This

None of this describes a blind spot SEO practitioners have missed. A capable SEO program already works on technical accessibility and information architecture, content quality, brand authority and reputation, outside citations, entities and structured information, user behavior and competitive performance. Some SEO consultants already examine how their clients appear in AI-generated answers.

The overlap with what’s sold as AI visibility is substantial, and it doesn’t make either kind of work illegitimate. An AI visibility consultancy can be doing plain SEO. An SEO agency can be doing exactly the examination described above. And some of what’s sold under either name is manipulation, which Google’s guide is right to warn against.

The label tells you very little. What matters is what the work investigates, what it measures and what it changes.

What Each Measurement Can Establish

Traditional SEO reporting is good at what it measures: rankings, impressions, clicks, traffic and conversions from search. Google’s new generative AI performance report in Search Console, rolled out to all sites at the end of August 2026, adds impressions for AI Overviews and AI Mode, counting how many times links to your site “were shown to a user in a generative AI feature.” That tells you your pages were linked. It doesn’t tell you what the answer said about your business, which competitors appeared alongside it, or which of them it favored.

AI visibility monitoring has limits of its own, and they deserve the same scrutiny. It runs defined questions and records what comes back: samples of possible answers, not the private conversations customers are having. A single result tells you what happened in one interaction, not how consistently the system will respond. SparkToro’s January 2026 research found less than a 1 in 100 chance that ChatGPT or Google’s AI, asked the same question 100 times, would return the same list of brands in any two responses. Repeated, carefully designed sampling is a different matter. Across well-chosen questions and many runs, it can reveal real patterns: which businesses appear consistently, how they’re described and against whom. Even then, a citation is not an endorsement, a mention is not a recommendation, and a recommendation is not a purchase.

Search reporting measures discoverability and performance within search. AI monitoring can examine how businesses are represented and compared across sampled conversations. Neither provides a complete picture of the decisions customers are making.

What Strong SEO Lets You Conclude

A company that has invested well in SEO and ranks strongly for the terms that matter to it can legitimately conclude a good deal. Its pages are accessible and performing well for relevant searches, and it has made information available that AI systems may also use. It is well placed to be found by systems built on search, though no ranking guarantees retrieval by any particular assistant.

What it can’t conclude from those results is how an assistant describes it when a buyer lists specific requirements, which competitors it is compared with, or whether a less prominent competitor with better-documented proof is being presented in its place. Those questions sit inside conversations the company can’t see.

For many businesses, the existing SEO program will be enough. Where customers buy on familiarity or proximity, purchases are simple and requirements rarely vary, there may be little to gain from looking further. The second question matters more where purchases are considered, requirements are specific, and buyers compare suppliers before making contact.

Much of AI visibility may indeed be SEO under another name. But good search performance doesn’t settle how a business will be represented or compared when an AI assistant helps a customer choose. Whatever we call the work, that remains a different question worth examining.