Your Competitor Didn’t Beat You on Google. They Won the Whole Category on ChatGPT
Quick summary: a Semrush study with Kevin Indig analyzed 1,094 categories and more than 50,000 brands on ChatGPT and found a clear owner in only 15.2% of them. The other 84.8% remain open, split between categories with an emerging leader and categories that are still fully undecided. The detail that changes the game for anyone doing brand audits: traditional SEO metrics, like organic traffic and Authority Score, only predict who wins the category in roughly half the cases. Losing to a competitor in AI happens for a different reason than losing position on Google, and you can check it in a few minutes.
What does “winning a category” mean in an AI’s answer?
Winning a single question isn’t the same as winning a category. Two related questions in the same buyer’s journey can surface completely different brands in ChatGPT’s answer. Gaining visibility on one prompt and disappearing on the next is the most common pattern found in the study, not the exception.
The study defines a category as a thematic grouping of five representative prompts: definition, comparison, alternative, use case, and purchase-decision question. The “owner” of a category is the brand mentioned in at least four of those five prompts, with an advantage of at least five percentage points over the runner-up. The question that matters is no longer “do I rank for this keyword,” it’s “when the AI answers any question within my topic, does my brand keep showing up?”
How many categories actually have an owner today?
Of the 1,094 U.S. categories analyzed between January and June 2026, 15.2% had a clear owner. Another 31.2% had an emerging leader, a brand appearing in at least three prompts without reaching the ownership threshold. And 53.7% remained fully undecided, with no brand appearing in three or more prompts.
Adding up emerging leader and undecided categories, 84.8% of the total remains up for grabs. For any agency or brand running an AI visibility diagnosis, that number is already the first point of reference: the odds that a specific client’s category is already locked up are, statistically, low.
Why doesn’t traditional SEO predict who wins on ChatGPT?
Because the set of factors that decides category ownership in AI only partially overlaps with the set that decides traditional ranking. Comparing the owner against the runner-up in each category, the owner had higher branded search volume in 55.7% of the pairs, higher organic traffic in 48.4%, and a higher Authority Score in just 52.5%. In nearly half the cases, the runner-up held at least one of those traditional advantages and still didn’t win the category.
The same disconnect shows up between citation and mention. Only 21% of the most-cited domains in a category were also the most-mentioned brand in the answer, and the correlation between the two metrics is slightly negative. That lines up with earlier Semrush research on “phantom citation,” a source that gets cited in the footnotes of an answer without becoming an actual brand mention. Strong SEO helps you get into the conversation. It doesn’t guarantee you’re the brand the AI names when someone asks “which one should I choose.”
A companion study from Growth Memo on how consumers navigate purchase decisions with AI helps explain why: a good share of people using AI to decide don’t even click through to a source, they just read the name the answer mentions first. That gap between citation and mention is worth paying close attention to. People who create content with GEO in mind tend to measure success by citation, showing up in the footnotes of an answer with a link. The data shows that’s necessary, but it isn’t the main prize. The prize is showing up in the body of the answer, as the name the AI actually recommends to the user, and fewer than a quarter of the most-cited domains in a category pull off that second result. Optimizing only for citation, without paying attention to how the brand is mentioned within the text of the answer, solves half the problem. It also explains why strong SEO alone doesn’t guarantee category ownership: it helps you get into the conversation, but a different set of factors decides who becomes the reference, factors tied more to how the brand is mentioned than to how much organic traffic it already had.
How do you check whether a client’s category already has an owner?
The process fits into a short work session. First, define the commercial category that matters to the client, not the isolated keyword: “accounting software for small business” is a category, not “best accounting software.”
Second, build five representative prompts covering the standard angles: a definition question (“what is X”), a comparison (“X or Y”), an alternative (“alternatives to X”), a use case (“best X for [specific situation]”), and a purchase decision (“which X should I choose”). Third, run the five prompts on ChatGPT and note down every brand mentioned in the answer, not just the source cited in the footnotes.
An example for the “project management tool” category: “what is a project management tool,” “Asana or Monday, which one should I choose,” “alternatives to Trello,” “best project management tool for a small marketing team,” and “which project management tool should I use.” Running these five questions and noting every brand that appears in the body of the answer, not just in the source footnotes, already delivers the full diagnosis for that category.
Fourth, classify the result. If a brand appears in four or five of the five answers with a clear advantage over the others, the category already has an owner, and it’s probably not your client. If no brand appears consistently in three or more answers, the category is open, and that’s exactly where content investment pays off fastest. Fifth, repeat this test month over month for the client’s most important categories. Position in a category with an owner rarely changes, but an open category shifts from month to month, and monitoring it frequently is what lets you move in before another brand locks in the advantage.
After finding out the category is open, what next?
Semrush recommends four moves after the diagnosis. Identify the thematic areas that actually matter to the client’s business, starting from what their buyer evaluates before deciding, not a generic list of industry topics. Analyze current topical authority using an AI visibility tool, or by manually repeating the five-prompt test, to honestly know where the brand already has signal and where it’s behind.
Focus effort on two or three categories at a time, because trying to build authority across everything at once dilutes the signal in each one. And build topical presence on three fronts at the same time: broad content coverage within that category’s questions, relevant mentions in the sources ChatGPT tends to cite for that topic, such as Reddit, review sites and trade publications, and strong entity signal.
That last point tends to be the most neglected of the three. Entity signal isn’t just a nice-looking page, it’s correct structured data, the brand name written the same way across every channel where the brand appears, and third-party citations that explicitly tie the brand to the topic it wants to be known for answering. A brand that changes its name between the website, the social profile, and press materials makes its own entity recognition harder, for both traditional search engines and AI systems trying to decide who to mention.
The diagnosis result changes the conversation with the client
If the test shows a category with an owner and a wide margin, the honest recommendation isn’t to promise head-on competition in three months. In the study, an owner with a comfortable margin held the lead in 9 out of 10 month-over-month comparisons, and the few position swaps that did happen were concentrated in categories where the initial margin was already narrow. Competing head-on against an entrenched owner is a long-term bet, or a reason to target a more specific slice of the category where no clear owner exists yet.
If the test shows an open category, the conversation changes shape. That’s the scenario where recommending content production, coverage in relevant sources, and reinforced entity signal tends to pay off faster, because there’s no accumulated competitor advantage to overcome, just a lack of sufficient signal from any brand so far.
The five-prompt test doesn’t replace strategy. It decides which strategy makes sense to propose, and it avoids the generic conversation of “let’s work on your AI presence” without knowing whether the category is already decided or not.
Summary: diagnosis before strategy
- Winning a single prompt isn’t the same as winning a category. You need to show up consistently in at least 4 of 5 related questions
- 15.2% of the categories analyzed have a clear owner, 84.8% remain open or fully undecided
- Traditional SEO metrics only predict category ownership in about half the cases
- The five-prompt test (definition, comparison, alternative, use case, purchase decision) shows in minutes whether a client’s category already has an owner
- An open category calls for focus on a few topics at a time, presence in the sources AI cites, and consistent entity signal, not broad, shallow coverage
Before promising any client an AI visibility strategy this quarter, run this test first. The diagnosis changes completely depending on whether the category already has an owner or not. To understand why this window of opportunity is shorter than it looks, it’s worth reading the full data on topical authority in AI and, if you’d like help running this diagnosis for your niche, check out the strategic SEO services.
Source: AI visibility is a topic-level game: A study of 50,000 brands in ChatGPT, Semrush, in partnership with Kevin Indig, Growth Memo.