A Semrush analysis of 1,094 ChatGPT categories found that only 15.2 percent have a brand the chatbot names consistently across related buyer questions, leaving close to 85 percent open. The categories drawing the heaviest AI search demand are the least likely to have that kind of owner, not the most likely, according to the study Search Engine Land covered July 21. That inverts the assumption most SEO teams import from Google, where high query volume usually means an incumbent has the category locked down.
Semrush split the 1,094 categories into two halves by AI search volume and checked ownership rates within each. Just 11.3 percent of the higher-demand half cleared the ownership bar, against 19 percent among the lower-demand half. That top half accounts for 98 percent of the AI search demand in the sample, meaning most of the volume in this dataset sits inside categories still open to a new entrant.
Semrush and Kevin Indig of Growth Memo set the bar for an owner this way: the brand holding the largest mention share, present in four of five representative prompts, and clear of the runner-up by at least five percentage points. Another 31.2 percent of categories had an emerging leader that showed up often enough but never built that lead. The remaining 53.7 percent were unsettled, with no single brand appearing in three or more prompts.
Traditional SEO strength turned out to be a weak predictor of who wins a ChatGPT category. Comparing owners against runners-up, owners had higher organic traffic in only 48.4 percent of pairs, worse than a coin flip. Owners led on Authority Score in 52.5 percent of pairs and on branded search volume in 55.7 percent of pairs; branded search volume was the lone figure to clear the bar for statistical significance, and even that link was modest.
Kevin Indig framed the finding with more restraint than the topline number invites. “Treat that as a hypothesis, not a finding,” he wrote. “What the data actually shows is: traditional SEO metrics aren’t enough to explain who owns a topic. While they play their role, there’s more to it.”
Ownership here was scored on how often a brand is named inside ChatGPT’s answer text, not on the links cited underneath, a choice built on Indig’s earlier finding that users picked the top-mentioned brand as their final choice 74 percent of the time. The two signals rarely line up: in just 21 percent of categories did the domain cited most often belong to the brand named most often, and the correlation between mentions and citations was slightly negative, at negative 0.229. Teams tracking citation counts as a proxy for ChatGPT visibility are measuring something that barely overlaps with what actually determines category ownership.
Ownership also proved sticky once established. Brands that qualified as clear owners held onto the top spot in 90.4 percent of the month-over-month checks Semrush ran, while the leading brand in emerging and unsettled categories flipped in 1,950 of the 5,470 comparisons. The gap came down to margin: leads that later collapsed averaged 1.3 percentage points, while leads that held averaged 2.9 points, more than double.
One disclosure belongs here. Semrush owns Search Engine Land, and Semrush also sells the AI Visibility Toolkit this study draws on. That ownership does not make the sample or the math wrong, but the data comes from a company with a commercial stake in selling AI-visibility tracking, and readers should weigh it accordingly.
The practical read for a team choosing where to spend effort on ChatGPT visibility is about margin, not just volume. A category with room to clear the five-point ownership threshold by two or three points is worth more than a crowded, high-demand topic where even the leader is stuck near a one-point edge, since narrow leads flip at roughly twice the rate wide ones hold. The highest-demand categories carry the biggest prize precisely because they are the least owned, but only for a brand that can build a lead durable enough to survive the next monthly measurement.
Search Engine Land reported these findings July 21, citing a Semrush study by Margarita Loktionova with Kevin Indig of Growth Memo published July 20.
