Quick read: Semrush analyzed 1,094 US topic categories in ChatGPT, tracked monthly from January through June 2026 across more than 220,000 domains, 50,000+ brands, and 600,000+ citations. It defined a category "owner" as the brand with the highest share of mentions, appearing in at least four of five related prompts, and leading the runner-up by at least five percentage points. Only 15.2% of categories had a clear owner by that definition (Semrush; reported by Search Engine Land). Traditional SEO strength did not reliably predict which brand ended up owning a category — domain-level metrics like Authority Score and organic traffic were higher for the owner in only about half of comparisons, with branded search volume the one factor that showed a statistically significant edge.
An owner emailed me last month asking, more or less, if it was already too late — whether the big national chains in her industry had already locked up ChatGPT the way they'd locked up the first page of Google a decade ago. I went and checked her category. Nobody had locked up anything. Nobody was even close.
That's not a one-off. It's the norm, and now there's a number attached to it.
What "owning a topic" in ChatGPT actually means
Semrush's study, done with Kevin Indig, didn't just ask "does ChatGPT ever mention this brand." It asked a harder, more useful question: across a cluster of related buyer questions in the same category — the kind of prompts an actual customer would type, not one cherry-picked query — does one brand show up consistently enough to be the default answer?
Their bar for "owner" was specific: appear in at least four of the five related prompts for a category, and beat the second-place brand by at least five percentage points of mention share. That's a meaningfully higher bar than "got mentioned once." It's closer to "if you asked ChatGPT this question five different ways, would you keep getting the same answer."
By that bar, only 15.2% of the 1,094 categories Semrush tracked had a real owner. The other roughly 85% didn't have anybody meeting that standard — not because no brand was good enough, but because no brand had built the kind of consistent, cross-prompt presence the bar requires.
That's the part worth sitting with. This isn't "AI search is crowded and the incumbents already won." It's "most of the field hasn't shown up in a way that counts yet."
Why this matters if you've been putting off AEO
If you've held off on Answer Engine Optimization (AEO) — the practice of getting cited inside AI-generated answers instead of just ranked on a results page — because it felt like showing up late to a party where the invitations already went out, this data says otherwise. For that same roughly 85% of categories, there's no incumbent to dislodge. There's just an open seat.
That seat doesn't stay open by default, though. Semrush also found that once a brand does clear that ownership bar with a real five-point lead, it holds first place in about 90% of month-over-month comparisons. Decisive wins tend to stick. It's the narrow, contested leads that keep changing hands. So "open" doesn't mean "permanently up for grabs" — it means the window to become the default answer is open right now, in a way it clearly won't be once someone else locks it down the way Google's first page got locked down years ago.
I ran a dog grooming salon for years before this, and the instinct to assume the biggest, most established name in a category has already won the space is a familiar one — it's exactly how a lot of small operators talk themselves out of trying at all. The data here says that instinct is usually wrong, for this specific game, right now.
What actually seems to move the needle
Here's the part that should reframe how you think about this. Semrush checked whether standard SEO strength — the stuff a lot of AEO pitches lean on — actually predicted who ended up owning a category. Mostly, it didn't.
Category owners had a higher Authority Score than the runner-up in only 52.5% of comparisons — barely better than a coin flip. Higher organic traffic showed up in only 48.4% of comparisons — a literal coin flip, and technically the runner-up won that one more often. The only signal that reached statistical significance was branded search volume: owners had it higher in 55.7% of comparisons.
Read plainly: being a bigger, more optimized website in the traditional SEO sense doesn't reliably buy you the top spot in ChatGPT's answers. What correlates is something closer to being a name people already specifically go looking for — which tracks with how these models seem to weigh real-world signal and mention consistency over raw page-rank-style authority. That's a genuinely different game than the one a lot of "AI SEO" pitches are selling, and it's good news for a business that can't out-spend a national competitor on content volume but can be a consistently accurate, well-described answer to the same handful of questions its actual customers ask.
The mechanics that seem to support that kind of consistency: accurate, specific service and location information that doesn't contradict itself across your site and your listings; content structured so a model can pull a direct answer instead of having to infer one; and enough of a real presence — reviews, mentions, a Google Business Profile that matches reality — that a model has more than one source agreeing on who you are and what you do.
What won't help
Chasing one lucky prompt. Getting mentioned once, in response to one specific question you happened to test, isn't ownership by Semrush's own bar — it's a single data point. The study's whole premise is that consistency across a cluster of related questions is what separates a real answer from a fluke.
Buying a bigger backlink package because "that's what SEO does." The data above says domain authority and traffic were close to coin-flip predictors. Pouring more budget into the metric that barely moved the needle isn't a strategy, it's a habit carried over from a different game.
Treating a proprietary AI-visibility score as the finish line. Vendor scores can tell you where you stand relative to their own model, but they can't tell you whether you specifically show up when a real customer asks — and they don't replace checking your own category the way Semrush's methodology did.
Waiting for "the right time." With a 90% month-over-month retention rate for decisive leads, the cost of waiting isn't zero — it's the risk that someone else claims the seat first and then keeps it.
FAQ
What does it mean for a brand to "own a topic" in ChatGPT?
Per Semrush's definition, it means the brand has the highest share of mentions across a cluster of related prompts in that category, shows up in at least four of five related prompts, and leads the second-place brand by at least five percentage points. It's a higher bar than being mentioned once.
Does my business need to be huge to own a category in AI search?
No — that's the core finding here. Traditional SEO strength (site authority, traffic volume) predicted category ownership only about half the time in Semrush's study, which is close to random. A smaller, specific, consistently accurate business has a real shot at a category a much bigger, more generic competitor hasn't claimed.
If most categories don't have an owner, does that mean AI search doesn't matter yet?
The opposite — it means there's more room to become the answer than most owners assume, and once a real leader emerges, they tend to keep the position. Waiting doesn't keep the door open; it just gives someone else more time to walk through it first.
How is this different from ranking well on Google?
Google ranking is about one page competing for one query. Topic ownership in ChatGPT, as Semrush measured it, is about whether a brand shows up consistently across a whole family of related questions a customer might ask in different ways. It's a related but distinct discipline from traditional search ranking, and it requires overlapping but not identical work.
Can I check whether my own category already has an owner?
You can run a simplified version of this yourself — ask an AI tool the handful of ways a real customer would phrase questions about your category and see whether the same business keeps coming up. We've written a plain-English version of that audit you can run in about fifteen minutes.
Is this the same thing as the "small local shop beating a national chain" argument?
It's the data behind it. A related post here makes the case that AI search doesn't reward size the way traditional rankings historically did; this study is the clearest quantification yet of why that's true — traditional strength just doesn't predict the outcome the way people assume.
Will this stay true, or will AI search eventually consolidate the way Google did?
Nobody can promise a percentage won't shift. But Semrush's month-over-month data already shows a pattern familiar from Google's early years: once a real leader emerges with a decisive gap, it tends to hold the position. The categories still open now are the ones most likely to consolidate first.
Checking where your business actually stands in AI-generated answers — not guessing, not buying a vendor's opaque score — is part of what getting found on Google and AI covers, before deciding whether it's worth doing anything about it.
Want a back office for your site?
Builds and Care are by application. Quality projects only — quoted at application.
Apply for a build slot →RELATED READS